]\gt ; VI .
The of the Mechanical of Heat , deduced from some xperiment performed with the ) of establishing the relation between the Units ; together with an Investigation into the for Heat of Water at different Temperatures .
E. H. GRIFFITHS , M.A. , Assistant Lec:tur'er , Colle.qe , Carnbridge .
by .
R. T. GLAZEBROOK , F.R. Received January 19 , \mdash ; Read February 16 , [ PLATES ] Page .
Iutroduction . . . . . .
361 Section I.\mdash ; General description . . .
367 II.\mdash ; The Calorimeter and Coil 373 III.\mdash ; The method of maintaining a constant temperature 374 \mdash ; The motol and its connections . . . . . .
378 V.\mdash ; The pressure in the surrounding space . . . .
380 VI.\mdash ; The method of maintaining a consiant E.M.F. 381 VII.\mdash ; The measurement of resistance . . . . .
388 VlII.\mdash ; The measurement of time 414 IX.\mdash ; The measurement of mass 415 X.\mdash ; The measurement of temperature 420 XI.\mdash ; The heat developed by the stirring . . .
433 XII.\mdash ; The gain or loss by radiation , \amp ; c. . . .
441 XIII.\mdash ; The method of a -experiment 456 \mdash ; The calculation of the results . . . .
477 \mdash ; Discussion of the results . . . . .
Appendix I.\mdash ; The determination of the " " Null Point\ldquo ; .
500 II.\mdash ; A form of gas regulator . . . . .
503 INTRODUCTION .
THE necessity for a re-determination of the value of the mechanical equivalent may not be obvious at first sight .
The classic determinations by JOULE have undergone but little alteration at the hands of succeeding observers , and the researches of ROWLAND ( 1879 ) into this matter were of such an exhaustive nature that there would appear to be little room left for further investigation .
It should , however , be NDCCCXCIII .
'J ' .
MR. E. H. ITHS ON THE OF remembered even JOTTLE'S later determinations differ by as much as 1 part in and that marvellous as is the agreement , amongst themselves , of the results obtained by OWLA N , they , since his method of investigation was unaltered throuo h\ldquo ; stand in need of confirmation by different methods of observation .
, ROLL , as far as I have been able to ascertain , stands practically alone in his conclusion that the heat of water diminishes as the temperature rises from to C. It is difficult to conceive of a more important investigation ( for the purposes of accurate physical measurements ) than the determination of ths capacity for heat of water different temperatures , and it is to me a matter of extreme surprise that greater efforts have not been made trace the variation ( if any ) in its value .
The science of calorimetry must be regarded as in its infancy so long as its fundamental unit is a ma.tter of doubt .
Other observers who have attempted to obtain the value of the mechanical equivalent , by means of the work done by an electric current , have been hampered by constant rplex ties as to absolute values of the electrical units adopted .
The science of electrical measurements has now arrived ab such a stage thab its units may be as sufficiently established , and , therefore , the time seems particularly appropriate for an enquiry into the relation between those units and the mechanical ones .
The difficulties of such an investigation are , of course , great , as is shown by ths divergence of results obtained by those Qervers who , in recent years , have adopted the electrical method .
One cause of inaccuracy has been present in all determinations I examined , viz. , the increase in temperature of the conductor above temperature of the medium in which it was placed , and the corsequent tion in its resistance .
ROWLAND writes as follows:\mdash ; ' ' There can be no doubt that experiments depending on the heating of a wire give too small a value of the equivalent , seeing that the temperature of the wire during heating must be higher than that of the water surrounding it , and hence more heat will be generated than there should be An account is given in the following pages of the manner in which this difficulty has been overcome , and I think it will be seen that this objection to the electrical method of investigation is now removed .
The difficulties with regard to measurement of temperature are not peculiar to the electrical method of investigation , and , therefore , I need not dwell upon them .
I would , however , venture to add my expressions of astonishment to tbose of ROWLAND so many enquirers have attached so little importance to this point ; many investiators , whose methods have otherwise been of a high order of accuracy , having contented themselves with the mercurial thermometer as a standard .
The importance * See Table 'Report .
Assoc 1892 .
'Proceedings American Academy , ' June , 1879 , p. 15.3 .
THE MECHANICAL EQUIVALENT OF HEAT .
of close attention to the thermometry is shown by the fact that the difference between the air and mercury readings , even at low temperatures , will more than account for a change of decrease into increase in the specitic heat of water as the ture rises .
An indirect result of my present investigation is to add there proof ( if is necessary ) to the of the nents of thermometers , and it is remarkable that so exact and a lnethod is not more rally applied to that most difficult subject , thermometry .
Another difficulty which meets any enquirer into calorime } Sllrements , is the uncertain nature of the thermal unit ordinarily adopted .
I iake it that the uIlit is usually delined as the " " mount of heat to 1aise unit mass of from to 1o C The obvious objections to this unit are that no one has more than a vague idea as to its itude , and that in choosing such a of temperature we have selected , without doubt , the most difficult of all tempel.aturcs ab which to ascertain it .
-Again , it is not improbable that ovel such a watel may exhibit greater changes its properties than it does ab other temperatures .
The consequence is tors , snch ROWLAND , are compelled to express their results in terms of a variable unit , the theoretical one being an unascertained and indefinite quantity .
To escape this difficulty I have taken as my ndard the quanti of heat required to raise unit tnass of water through 1o C. of the airthermometer at C. from ) ; in other words , I assun1e the specific heat of water at C. to be 1 , and I venbure to suggest that of a better ) the above definition be accepted as that of the thermal u1li I have not included in this paper any prolonged analysis of the of previous observers .
ROWLAND has given a most complete summary , with references , of all results anterior to 1880 , the agreement between later observers is not bufficient to warrant any modification of his conclusions .
The main portion of this communication is devoted to an account of the experiments performed in my laboratory during 1892 , and , in order to avoid the introduction of redundant matter that account , I here give a brief summary of my previous investigations .
I commenced work in 1887 , and the general principle on which proposed to proceed was that of eliminating the ects of conduction , tion , \amp ; c. , rather of ascertaining the loss or gain due to such causes .
If a calorimetel is suspended in a chamber whose walls are kept at a constant temperature , it is obvions that if the initial temperature of the calorimeter is below that of the extel.nal envelope , and if the ] orimeter has its temperature steadily raised by means of * Personally I sbould have to scIect a higher temperature , since a tempel a at of our ordinary surroundings is , at all times , more easily obtainable than a lowel one ; but as C. hos become so commonly accepted as a standard of temperature in cloctrical measurements , I have it as my standaId .
gives all results published beforc , 1892 .
'Annales do Chimie ct dc Physique , ' vol. 27 , p. fl. GRIFFITHS VALUE , a time will come when the loss by radiation , \amp ; c. , must equal the gain ienced by the ca]orimeter during the time it was below the temperature of the surrounding space .
The temperature at which the effects of \amp ; c. , are eliminated , I term the\ldquo ; null point and an investigation into its position indicates that if be the time that the calorinleter was beneath the external temperature , and the time required to raise it from its initial temperature to the number of degrees above , that it was previously beneath , the temperature of the surrounding envelope , then the would arl .
iv at its null point at a time somewhere between and In the Appendix will be found an investigation into the exact position of this point .
1887-1889.\mdash ; During these years my whole attention was ected to an effort to ascertain the heat developed by the current while the calorimeter was passing from its initial temperature to this null point .
Attractive as the method appeared , I ultimately relinquished it , for I wished to direct my attention as much to the investigation of the changes in the specific heat of water as to the determination of the mechanical equivalent , and , although the method is an admirable one for the latter , ib is ot so suitable for the purposes of the former enquiry .
My general method of conducting the experiments was the same as that observed in 1892 and described in the succeeding pages .
From the outset , I had determined to vary all the conditions as much as possible , believing that it was only by such means that constant sources of error could be detected .
The agreement , amongst individual experiments taken under the same conditions was , if anything , at times more marked in the experiments of 1892 ; nevertheless , when the final reduction of the results took place , fatal discrepancies invariably showed themselves .
For example , the water equivalent , as deduced from the experiments , increased when the mass of the water was increased , and experiments conducted with a high electromotive force invariably gave too a value for the time as compared with that obtained when a lower electrolnotive was used .
These discrepancies , as it now appears , may be attributed to two causes:\mdash ; ( 1 .
) To insufficient stirring ; ( 2 .
) To t , he rise in temperature of the wire , preyiously referred to .
Throughout these experiments I was conscious of the error introduced by ignorance of bhe actual temperature of the wire , but it was not until the summer of 1892 that a satisfactory method of estimating this difference of temperature was adopted .
Unfortunately , many of my experiments in years were conducted with wires coated with a thick insulating covering , and it is , therefore , impossible to apply the correction obtained this year to the reduction of the earlier results .
The observations were excellent in themselves , and some hundreds of experiments were performed .
Thoy all give too high a value of ( almost invariably above ) , and are of no -alue except for the experience gained by .
means and the improvements they suggested in the ) THE MECHANICAL VALENT OF HEAT .
Again , the thelmometry was not of a sufficiently satisfactory nature .
The thermometers were standardized by comparison with B.A. ated at Kew , * As these B.A. thermometers were only divided into of a degl.ee , the of a degree had be estimated , and the ordel of accuracy was not 1890.\mdash ; Throughout this year I devoted my attention entirely to the nent of temperatures by platinum thermometers , and , at that time , proposed to use such thermometers in place of mercurial ones during my experiments .
Excellent , however , as platinum mometers are , .
the accurate determination of temperature , they not suitable when the obsel .
Vations have to be taken at exact intervals of time .
A whose readings are at all times visible , must be used in such a case .
1891.\mdash ; By the kind permission of the Mastel and Fellows of Sidney College , the Chemical Laboratory was placed at my disposal during this summer .
This building was unsuitable for physical work , the traffic in the adjoining was so great that galvanometers became at times unmanageable , and many of the observations had to be taken during the night , under conditions somewhat conducive to pel.sonal The steel regulating chamber , described in Section III .
, was used in this yeal ( 1891 ) , and the results are therefore of more value than the ones ; but the same persistent error in the measurement of still emained , and the form of stirrer adopted was unsatisfactory .
The summer 's work was , howevel , of some value , as it gave sufficient data for calibrating the mercury thelmometer when rising .
A platinum wire coil , having a single coating of amber varnish , was used , and it is reasonable to suppose that its increase of tenlpelature above the sul.rounding water was of a similar order to that ascertained to be the case in the coil of 1892 .
Assuming the increase of its esistance to bear the same ratio to its total esistance as in the case of the 1892 coil , and , applying a similar COl.rectioI ) to otll results , we obtain expressed in terms of water at , and assuming the coefficient of the change in specific heat of water obtained in 1892 , we get 4 .
as the value expressed in of the thermal unit as previously defined .
The accm.acy of the Kew corrections for the inequality of the bore has been eviously discussed , see ' B.A. Report , ' 1890 .
I find that many observers , by means of such , determine ( without using a micrometer scale ) to of a degree .
I envy , but cannot claim to , such powers of observation .
The wires connecting the lid of the imeter with the steel chamber had ( in 1891 ) a slightly higher resistance than those used in 1892 .
I am unable to make the necessary corl tion , as I have not sufficient data .
effect of the ection which less than in 5000 ) would bo to slightly tho value of here givcll .
This agrees within 1 in 2000 with our later determinations .
* 1892.\mdash ; The difficulties experienced in obtaining an altogether suitable room for the work compelled me to have a pecial building erected in which to carry on the investigation .
The appal tus is so involved and complicated that it requires weeks to put it together and to get it into working order , and such spare time as I had in the early of this year was devoted to these preparations .
On previous occasions the constant day and night changes in temperature of the rooms in which experiments conducted had been a cause of uncertainty ; for the numbers , obtained by using boxes Clark cells whose temperature is constantly changing , are always of doubtful accuracy .
I therefore designed a regulator which , except in cases of severe frost or extreme heat , automatically maintained the room at a constant temperature .
I found .
this arrangement so effective that I have , in the hope that it may be useful to others , given a short description of it in Appendix II .
If I endeavoured to adequately describe how much I owe to the assistance of others during this investigation I should unduly crease the length of this introduction .
I must , however , express my thanks to the Council of the Royal Society , to Lord , Lord RAYLEIGH , Professor G. DARWIN , Professor HICKS , Professor J. J. THOMSON , Dr. FEUSZNER , Dr. GUILLAUME , Mr. CA LLENDAR , Mr. VERNON HARCOURT , Mr. HEYCOCK , .
LARMOR , and Mr. NEVILLE for the ement and assistance they have afforded me .
More especially am I indebted to Mr. GLAZEBROOK , not only for his advice and help , but for the careful and laborious comparison of my l.esistance box with the standards of the B.A. ; and also to Mr. SKINNER for his repeated comparison of my Clark cells with the Cavendish standards .
My thanks are also due to Mr. A. IVATT , B.A. , Christ 's College , Cambridge , for his assistance in the experimental work during 1887 and part of 1888 .
Nearly the whole of the apparatus ( some of it of a novel and intricate character ) was constructed by , or under the direction of , Mr. F. THOMAS , Jesus Lane , Cambridge .
Had it not been for his unremitting attention and skill , the results of my investigation would have been far from satisfactory .
From 1888 onwards I was assisted by Mr. G. M. CLARK , B.A. , Sidney College , Cambridge , who has from the commencement of our joint work been indefatigable in his exertions .
Our respective contributions are so intimately associated that I find it impossible to make the remainder of communication in the first person , and although , by his own wish , his name does not appear on the title-page , it should , in justice , be regarded as a joint contribution .
* The Clal k cells used as our standards of given in Table .
of GtAZr .
BROOK and papel ; ' Phil. Trans , 1892 , p. 605 .
THE MECHANICAL EQUIVALENT OF HEAT .
SECTION I.\mdash ; GENERAL DESCRIPTION . .
Let be the resistance of a at a certain standard temperature , then if the ends of the conductor be kept at a constant difference of potential ( 1 ) , where is the quantity of heat generated by the current , is time , and the meohanical equivalent of heat .
If the wire is immersed in water , and if the capacity for heat of the water and the calorimeter at the standard temperature is , then .
( 2 ) .
It to distribute the heat generated in the wire throughout the whole calorimeter as quickly as possible , otherwise we cannot accurately observe hence , since work must be done when stirring , we have to deal with a mechanical as well as with an electrical supply .
Let the rise in temperature per second due to the mechanical supply be , and let denote the quantity of heat generated by both electrical and mechanical sources .
Then .
( 3 ) .
It is certain that some heat will be gained or lost by radiation , conduction , and convection .
Let be the rise or fall per second in temperature due to radiation , \amp ; c. , when the difference between the external and internal temperatures is 1o C. Denote the temperature of the surrounding envelope by , and the temperature of the calorimeter at any time by ; also suppose that becomes and becomes when becomes We now have ( 4 ) , where denotes the gain of heat due to all causes , hence ( 5 ) .
* It is shown in Section .
that we are justified in assuming that law of cooling holds true over our range of temperatul.e .
If , this becomes ( 6 ) .
Provided that the supply is regular , and is kept constant , we can , by a sufficient number of observations over small ranges , determine the values of for different values of , and by substituting in equation 5 , we obtain the value of , where A is known .
Assuming that over small anges the values of and are linear functions of we have , and hence M. ( 7 ) , Now the values of and at the standard temperature can be ascertained by isons with the stalldards , and the value of can be ascertained by direct measurements of at different temperatures , hence .
M. , where is known .
( 8 ) , we have thus one equation connecting the three quantities , , and Let and be the capacities for heat of the water and the calorimeter respecat the standard temperature , and let the temperature coefficients of their specific heats be and respectively , then hen ce equation ( 8 ) becomes . . .
( 9 ) .
If observations are taken with different weights and ) of water , we obtain the corresponding values of , thus and ; * this , and similar cases , we use the suffix to denote the sources of heat .
The value of , during our expel'iments , was about C. true value of where is the excess of the temperature of wire above tlJe temperature of the calorimeter .
See Section XIV .
, p. 478 : EQUIVALENT by subtraction we obtain . . .
( 10 ) .
Hence , when , the value of can be found without .
the value of the water equivalent or the temperature coefficient of the specific heat of water .
As is a constant quantity we can , by repeating the observations at different temperatures , obtain from equation ( 10 ) the value of Or we can obtain without finding , for repeating the observations at temperatures and and the second suffix of denoting the temperatures .
Hence , by division , we obtain ( 11 ) .
An alternative method of finding is to first find by equation ( 11 ) , and then to find the value of and The value of can then be deduced from a single experiment by means of equations ( s ) , and as this method enables us more easily to compare the results of individual experiments we have , as a rule , adopted it in our reductions .
In the remainder of this section we briefly indicate the manner in which we have determined the various quantities grouped together and denoted in the above equations by B. The calorimeter was suspended by means of glass tubes in an air-tight chamber whose walls were kept at a constant temperature .
The pressure of the dry air in this chamber was reduced to under millim. The water in the calorimeter was stirred at a very rapid rate throughout the experiments , and the value of ascertained at all parts of the range to C. ; firstly , when the work was done by the stirrer only , secondly , when the rise was due to both the electrical and the mechanical supply of heat .
The masses of water used during the experiments varied in the ratio of about 1 to 3 , and the difference of potential at the ends of the coil was so altered that the heat developed by the current was changed in the proportion of 1 to 9 .
As shown by equation ( 10 ) , there was no necessity to ascertain the water equivalent of the calorimeter , although it was found convenient , as a check upon the These values can be obtained from ths preceding equabions without first determining J. The process is fully explained in Section XIV .
MIJCCCXCIII .
calculations , to deduce it from the differences in time caused by differences in mass of the contained water .
The walls of the chamber containing the calorimeter were maintained at a constant temperature throughout each experiment , this being , in general , the mean temperature of the above-mentioned range .
Direct observations of the changes in caused by changes in , enabled us to ascertain the values of both and ( 1 .
) Potential ( E).\mdash ; A detailed account is given in Section VI .
of the method by which the ends of the calorimeter coil were maintained at a constant potential difference .
Briefly , the principle on which our method depended was as follows .
The extremities of the wires , will ( when the galvanometer shows no deflection ) be maintained at a difference of potential equal to that due to the cells at Let be an adjustable resistance placed in the circuit which communicates with ( the storage cells ) , and let be the resistance of the wire .
Whatever variations may take place in and in the E.M.F. at it is always possible ( provided the adjustment of is sufficiently under control ) to maintain , by close attention to the indjcations of the galvanometer , the points and at a constant difference of potential .
A special rheosLat was designed by means of which it was found possible , in spite of variations in the resistance of , to maintain the potential-difference unchanged throughout an experiment , and we believe that in no case did the variations exceed oo of the mean difference of potential during each experiment .
The Clark cells ( which were placed at ) were constructed by us according to the directions of Messrs. GLAZEBROOK and SKINNER , and have , on several occasions , been directly compared with the Cavendish and indirectly with the Berlin standard .
Their differences from the standard are small , and their mean E.M.F. at C. volts ( see Section VI ( 2 ) .\mdash ; Had it been possible to maintain a constant value for it would have greatly simplified both the experimental work and the calculations .
In the year 1890 we devoted much time to the examination of the various copper-manganeSe nickel alloys , and we performed a series of determiuations with a coil whose tempera THE lsfECHANICAL EQUIVALENT OF 3 .
71 ient was practically zero .
The reasons which led us to reject these alloys and adopt a platmum wire will , we think , be found sufficient see VII Tho value of was first determined by a dial-box legal ohms Constructed by Messrs. ELLIOTT .
* .
GLAZEBROOK has been so kin as to perform a complete standardization of this box by meanS of the B.A. Standards .
The correctionS have been applied , and the valueS of expressed in true ohms , as defined by the B.A. Report , lS92 .
The method by which we ascertained the true temperature , and therefore , the resistance of the wire when traversed by a current , is fully described We found that the difference in temperature between the and that mdicated by the thermometer could be determined , and , other conditions bemg equal , varied as The arrangement of our connections enabled us to dispense with exterior resistance , as the ends of the coil itself were at a constant potential-difference .
3 .
) Current ( C).\mdash ; The cUrrent was maintamed by storage cells of the ( L ) type , supplied to us by Messrs. BAILEY and GRUNDY .
The E.M.F. of these cells was constant .
This constancy , although not a necessity , was a As pomted out , was not directly measured .
4 .
Time ( T ) \mdash ; An electrical clock with a seconds pendulum was used as our standard for time .
It was compared at intervals with a chronometer by .
A chronograph was conbrolled by this clock .
The rate of the dock was a losing one until August 21st , but after that date its error was less than 1/ 25,000 , and no correction was necessary .
5 .
balance nsitive to a change of 1/ 100,000 of the least mass measured by us , and a set of were used in our determination of mass .
( 6 .
) Temperature ( ) .\mdash ; At the time of writing December , 1892 our thermometry is based on measurements made by platinum-thermometers .
We propose to make , at an date , a direct comparison of our standard thermometer with the air thermometer by means of the apparatus described Mr. CALLENDAR , has been so kind as to promise his assistance .
In a previous paperSthe details of a eful comparison of the platinum and the airthermometer have been given .
It was then shown that the platinum-air difference Curve gave values of at all temperatures from to C. , accurate within of the real value of , and that discrepancies appeared to have an experimental experience of two years ' work with platinum-thermometers but increased our confidence in them Should a direct comparison with the of have been given in a previous paper .
'Phil .
Trans A. , 1891 , p. 44 .
were -standardized by Messrs. OERTLING in August , 1892 .
Roy .
Jan. , 1891 .
S ' Phil. Trans 1891 , sophical Magazine , ' Decembel , 1892 .
air-thermometer modify our determinations of differences of temperature ( and it is only differences which are important to us ) , our results will be modified accordingly .
This will not , owever , necessitate any repetition of the experimental work , as the corrections involved will be of a numerical order only .
Measurement of the Heat by Stirring .
The difliculties that we have with in this portion of our work have been very great .
As will be shown in Section XI .
, we have , against our will , been compeled to increase the work done by the stirrer until the heat thus generated became about 1/ 10 of the whole supply .
We found continuous and very rapid stirring to be a necessity .
Variations in and in the thermometer readings and in the water equivalent of the calorimeter invariably followed if the mixing was insufficient .
Commencing in 1888 with a stirrer revolving at the rate of about 100 revolutions per minute , the rate was gradually increased until , in our last series , we aimed ( with , it is true , a different form of stirrer ) at a rate of nearly 2000 revolutions per minute .
The accurate measurement of the heat thus generated is only second in importance to the determination of the heat due to the current , and , accordingly , a great portion of our time and attention was directed to it .
Fortunately , we were able to completely establish the law that , with our form of stirrer , the work done varied directly as being the rate of rotation .
From this time our difficulties rapidly diminished .
Radiation , Conduction , ection , We endeavoured , as far as possible , to diminish loss or gain due to the above causes .
has shown that there is a rapid fall in the rate of loss or gain by radiation and convection when the pressure of the surrounding gas falls below 1 millim. Most of our observations were taken with the surrounding pressure under .
millim. The calorimeter itself had a bright polished surface , and the walls of the enclosing chamber were clothed with bright speculum metal .
In Section XI .
, we show that throughout our raIlge of temperature , NEWTON'S law of radiation holds -i.e .
, the loss or gain ries directly as the differences of temperature .
It must , however , be remembered that we have not endeavoured to separate the partial effects of the radiation , conduction , and convection , and hence we can only say that the combined effect obeys this law .
The total loss or gain , due to the above causes , was determined separately for each change in the contained mass .
Any error in this part of our work would have but a small effect on the mean value of , since , in each experiment ( if taken as a whole ) , the total loss or gain by conduction , convection , and radiation , must have been small .
* ' Phil. Trans 1887 , , p. 445 .
THE HANICAL EQUIVALENT OF HEAT .
SECTION \mdash ; THE CALORIMETER AND COIL .
( See Plate 2 , fig. 2 .
) The calorimeter , which was of a cylindrical form , was made of ' ' gilding metal\ldquo ; \mdash ; 73 per cent. copper and 27 per cent. zinc .
The weight of the lower part was grms. ; of the lid grms. The internal diameter and the depth were each 8 centims. Its capacity was therefore about 400 cub. centims. The weight of the calorimeter may appear excessive for this capacity .
However , as any " " panting\ldquo ; of the lid or base would have thrown the stirrer out of adjustment and as the pressure on the flat of the calorimeter would amount to about , when the space surrounding it was vacuous , it was evident that considerable thickness was necessary to secure rigidity .
The two parts of the calorimeter were held together by eight screw bolts ; a lead ring previously placed between the parts was crushed by the tightening of the screws and thus the join was rendered airtight .
It was therefore necessary to make both the lid and the projecting lip of the calorimeter sufficiently strong to bear the strain without bending .
As we expected to determine the water equivalent with as much accuracy as other quantities , the large mass of the neter did not appear to be a serious objection .
The whole of the calorimeter was heavily gilded both inside and out , and the surfaces polished .
The lead ring was covered with gold leaf because considerable quantities of were formed if the water had access to it ; thus , no metal except gold was exposed to the action of the water .
An advantage of the gold exterior was that its surface did not tarnish , ) thus its coefficient of radiation remained constant .
Fixed on the interior of the lid were eight metal tubes , 1 centim. in length .
Three of these were 18 millims. in diameter and served to hold the glass tubes by which the calorimeter was suspended .
The others were 5 millims. in diameter and supported the glass rods on which the coil was wound .
These rods reached to within a couple of millimeters of the base when the lid was placed in position .
Two copper rods , centims. in length ( 14 B.W.G. ) passed from the insulated junctions at and ( fig. 2 , p. 382 ) to nearly the bottom of the calorimeter .
These were also heavily gilded , and the ends of the platinum wires were fastened to their lower extremities .
The last coil used by us was about 13 inches in length , and was would round the lower end of the glass rods in such a manner , that the difference of potential of any two adjacent points could never be great .
Thus the first 6 inches were wound on one side , the remainder on the further side of the calorimeter .
A depth of centim. of water completely covered the whole coil .
The glass tubes passing through the lid were held in position by perforated indiarubber corks .
Great pressure had to be exerted to force the tubes ' home and $ A pure gold ring was tried , but it was found impossible to crush it sufficiently to render the join air-tight .
MR. E. H. RIFFITHS ON THE VALUEi OF ' several were broken in the process .
The external diameter of the glass tubes millims. , the interior diameter of the gold tubes 18 millims. ; thus the of india-rubber was very thin .
The lower ends of the glass tubes were turned outwards so as to nearly cover the lower end of the india-rubber , of which the small exposed surface was originally coated with gutta-percha ; but ( as explained elsewhere ) this was removed after , and replaced by amber varnish .
* The india-rubber corks , where they appeared above the lid , were , from the commencement , coated with amber varnish .
No diffusion appeared to take place through these corks , and , as other observers seem to have been troubled by such diffusion when using india-rubber , it is probable that the coating of amber varnish ( which adheres very firmly to the rubber ) is an effective remedy .
In any case it prevented any contact between the rubber and the contained water .
We may thus assert that the vRter came into contact with the following substances only\mdash ; gold , amber , glass , agate , gutta-percha ( before experiment ) and ( after experiment ) some hard shellac , very small quantities of which were used to prevent slippin of the coil on the glass rods , the substance previously used for this purpose being gutta-percha .
The platinum coil itself was coated with amber varnish , but it is possible , that owing to cracks in its coating caused by movements of the wire , the water had access to it in places .
The small slabs of ebonite used at the junctions and ( fig. 4 , p. 396 ) were also covered with several layers of the same invaluable varnish .
SECTION III.\mdash ; THE METHOD 0F MAINTAINING THE WALLS SURROUNDING THE CALORIMETER AT A CONSTANT TOMPERATURE .
The necessity of maintaining at a constant temperature the walls of the chamber containing the calorimeter is so obvious that it seems unnecessary to dwell upon it .
In our earlier experiments , the calorimeter was suspended in a copper vessel which was immersed in a large tank , and the uniformity of the temperature depended on the of the observer in adjusting the temperature of the tank water by altering the supply of hot and cold water admitted .
The method adopted by us in 1891 and 1892 worked automatically , and enabled to main tain a chamber at any temperature within the range of our experiments , whether that temperature was above or below that of the room .
iefly , the arrangement may be described as an thermometer-bulb within which was the calorimeter .
A section and plan , on a scale of , are shown on Plate 3 .
The steel chamber , ( a very fine piece of work ) was constructed by Messrs. WHITWORTH and Co. A detailed description of its various parts is unnecessary , as all particulars are given in the plate referred to .
The annular space ( coloured black ) was filled with mercury .
Note , p. 392 , infra .
The shellac was melt on , and not deposited from its solution .
A hard surface is rarely obtained by the latter method .
THE MECHANICAL EQUIVALENT OF HEAT .
As it was advisable to free that space from air bubbles , the screw plug was removed before the insertion of the mercury , and replaced by a tube leading to a GEISSLER'S mercury pump .
A tube from dipped into a basin of hot mercury , and in this tube was a tap which was at first closed .
The .
in the annular space was withdrawn until the pressure was a small fraction of a millimeter , and the hot mercury was then slowly admitted\mdash ; the exhaust pump being worked at intervals .
When the mercury became visible within the tube at , the tube was cautiously removed , and the airtight screw plug replaced .
The mercury necessary to fill the space weighed So true were the surfaces of the two parts of the metal chamber , which were screwed together by the bolts at and , that , although no packing except the finest possible coating of grease had been placed upon them , they proved absolutely airtight .
These two portions of the chamber were each formed from a single block of steel and , when placed together , somewhat resembled a double hat-box .
The inner and outer surfaces were turned smooth and true , and then were nickel-plated .
Two holes on opposite sides were bored down between the mercury chamber and the inner surface , as shown at and .
These holes were filled with mercury and , their inner walls being very thin , the thermometer placed in them would accurately indicate the temperature of the inner surface of the chamber .
This inner surface , as also the base of the chamber , was , in our more recent experiments , covered with highly-polished speculum metal .
A ring of lead wire was placed round the top edge of the chamber , on which was laid that portion of the lid shown in section at and , and in plan at and , and the twelve bolts securing it were screwed home .
Many as have been our difficulties with regard to air-tight joints , those made in this manner , viz. , by the crushing of a lead ring between two surfaces , have never given us any anxiety .
The second lid , from which the calorimeter was suspended , and whose section is shown at , and plan at , was secured in a similar.manner .
The necessity for this double lid is not , at first sight , obvious , but it must be remembered that it was impossible to separate the calorimeter from the portion of the lid to which it was attached , and , therefore , a heavy steel plate would have been inconvenient , and have rendered direct weighings impossible .
On the other hand , the total pressure on this lid , when the chamber was exhausted , was very great , and , had the whole of it consisted of one thin piece , it is probable that it would not have withstood the strain without 'panting and , as it was necessary on account of the stirring rod , \amp ; c. , that the whole apparatus should be absolutely rigid , it was essential to avoid any phenomenon of this kind .
The first lid , therefore , is of considerable thickness , and the orifice in it only just sufficient to allow for the introduction of the calorimeter .
Again , if the lid had been of one piece , it would have been impossible to remove it without disturbing the tubes and The tube communicated with a five-way Sprengel pump and a McLeod gauge ; the tube with a Geissler pump and a water pump .
The arrangements for drying and exhausting by means of these tubes is described in Section V. S76 MR. E. H. G.RIFFITHS ON THE VALUE OF The steel chambel , together with the mercury , weighed nearly 2 cwts .
It stood on small tripo in a galvanized iron tank , whose capacity was about 20 gallons .
The itself was placed on a bed of Portland cement before the cement had hardened , thus the whole erection was firm and stable .
When the outer tank was filled , surface of the water was about 3 inches above the lid of the steel chamber and ltained at that level by a constant inflow of water , which entered almost the screw and left by an overflow pipe shown at .
The screw at rotated such a direction that the water was forced downwards , passed under the chamber , and returned across the lid .
screw revolved about 800 times a and the stirring was very thorough , so that it was impossible f , o dotect any of temperature in different portiorJs of the tank , even when hot water was admitted , unless the thermometer was held immediately against the entrance A glass tube with narrow bore ( Sprengel tubing ) was fixed into the steel tube at and communicated with the regulating apparatus shown in Plate 4 , fig. 1 .
is regulator\mdash ; the gas entering at and leaving at .
The tube as about 1 inch diameter and was drawn to a fine opening at its lower end .
The thin glass passed down the centre of this tube and projected at its lowest extremity into narrow opening .
To render the partition air-tight , its edges were coated with shellac , and after it had been thrust into position , the tube was warmed until shellac melted .
The top of the tube was closed with shellac , thus the gas , at , could only arrive at by passing round the lower end of this partition , and very small movement of the mercury in the narrow neck sufficed to cut off the supply .
We adopted this form of regulator after considerable experience of the ordinary in which the supply passes down an inner tube and up the annular space around which under the defect that the adhesion between the mercury and inner tube causes the gas to remain cut off for some time after the mercury has menced to contract .
The form here shown is free from this defect , as it is found the mercury does not adhere to the thin glass partition .
Beneath the narrow a small reservoir ( capacity about 3 cub. centims. ) at allowed considerable to take place , and yet permitted the mercury to be kept in view .
A side tube from to the three-way taps and , and thus , by means of the tap at could be withdrawn when necessary .
The tubes and were about 4 in length\mdash ; K terminating in a thistle-funnel , in which a supply of mercury placed .
The tube was connected with the three-way tap , by which communicati could be made with the exhaust water-pump , or with the open air .
the was fixed a vertical scale of millimeters , and the whole of the tubing was wide-bore .
Through the tap at mercury from the arm could passed into the arm to any convenient height .
By means of the taps and any portion of this mercury could be passed into the regulator , and , if necessary , same amount could at any time be withdrawn by connecting with the exhaust THE MECHANICAL EQUIVAL ] ] OF HLAT .
means oftap ) and opening and H\mdash ; independently of the leyel of the merCLlry in the regulator .
This contrivance , therefore , rendered it possible to " " set\ldquo ; the instrument so as to produce any required change of temperature in the tank ; for it was found that the transference of about 312 millims. of the column into the regulator caused a difference of C. in the temperature of the tank .
It was thus possible to adjust the temperature to a very small fraction of a degree .
The connecting tube from to was bent into a spring of sufficient length to allow of some play to the lower end , and thus it was possible to withdraw the end from without moving the whole edifice .
The glass work was supported on a vertical board , about 5 feet in height , and the various taps were placed within easy reach of the observer .
All gas admitted at was dried by passing through tubes containing quicklime , as it was found that the regulator worked in a much more satisfactory manner when the surface of the mercury was clean .
After leaving , the gas entered a tube about 2 feet long , containing a very large number of small pin-holes .
The tube was connected by two cross pieces containing similar small holes , with two pilot lights fed by an independent supply .
A change of of a degree in tank visibly affected the brilliancy of the tiny gas jets , and a change of less made all the difference between totnl eclipse and a complete series of bright blue jets .
The supply of gas through the regulator was , however , sufficient when the mercury had receded far enough to maintain the jets at a length of nearly inch .
Placed above this row ofjets , but so far from the lights as to be unaffected by them , was a flat silver tube , the section of whose bore was a rectangle of about 1 inch wide and inch deep , and , through this tube , the water passed on its way to the tank .
The supply was regulated by a small ' constant level\ldquo ; tank of the usual pattern , suspended from the ceiling by a string .
Alterations in the level of this tank afforded a fine adjustment\mdash ; the temperature depending on the rate of flow .
The water was supplied direct from the main , and .
hence , even in the hottest summer weather , it was possible to maintain the tank at a temperature as low as ; in winter , as low as or C. Any temperature above this minimum was , of course , obtainable , but we considered it unadvisable to raise it above C. on account of the ebonite , \amp ; c. This regulator proved itself to be a most satisfactory instrument .
Although , when set to give a new temperature , it was slow in its action ( the capacity for heat of the tank and its conten being great ) it was none the less sure , and , after about half-an-hour from the readjustment , the temperature would be found bo have become steady .
If set to give a higher temperature , the thermometer at ( Plate 3 ) would pass that temperature by about a degree , and , after this preliminary oscillation , would become steady , although it is probable that this first followed by other oscillations , which were too small be apparent .
It is possible that the temperature of the inner surface of the steel chamber was , to some extent , lnner sir ace affected by the temperature of the calorimeter .
The change must , however , have been very small , since it was insufficient to affect the thermometer at .
It must NDCCCXCIII .
be remembered that any such change would at once produce an effect on mercury , and , by its action on the regulator , reproduce the original condition .
SECTION MOTOR AND ITS CONNECTIONS .
A belt passed from the motor to a vertical shaft fixed near the back of the This shaft carried " " speed-wheels , \ldquo ; from which ran two cords , one to the stirrer in outer tank , the other to the stirrer in the calorimeter .
In order to prevent any " " slipping with which we were at one time much troubled , all these cords through loose pulleys hung by strings , which , passing over small fixed pulleys , were connected with pans containing shot .
The tension of the cords could thus be regu .
lated by the weight of shot , and any change in their length , due to stretching , \amp ; c. , was of no consequence .
The shaft of the calorimeter stirrer was connected with the revolving piece above it by a double HOOKE'S joint .
Until experiment a short piece of brass tubing with a cross wire at each end served as the HOOKE'S joint , with the exception of and 18 , when only a single HOOKE'sjoint was used .
A difference in the " " stirring supply\ldquo ; was shown in these two periments ( see p. 450 ) , probably due to some difference in pressure between the agate cylinder at the bottom of the stirring-rod and the ring surrounding it .
After experiment the connection was made by a small circle of thin wire , which passed through a hole in the revolving shaft and dropped into a slit in the top of the stirring-rod .
Thus all rigidity in the connection was avoided , and the stirring effect was more regular after the change .
Above the stirring-shaft was a counter , whose axis carried at its upper extremity a V-pulley , and at its lower extremity the HooKE'Sjoint ( see Plate 2 , fig. 2 ) .
These counters move in steps , and not with a continuous sliding motion , and , as it was essential that should be able to read with considerable accuracy the time of each 1000 revolutions , this form of was the best adapted to the purpose .
The wear that these instruments stand is surprising , for , on many occasions , nearly a quarter of a million revolutions per day were recorded , and the same counter was used by us throughout our ments of 1891 and 1892 , with the exception of experiments to 34 , when original counter , having shown some slight signs of wear , was replaced by another .
The motor ( one of the fan type ) was by BAILEY and Co. , and of the usual working at the Sidney Laboratory in 1891 , the changes in pressure , and fore in he speed of the motor , were so frequent that we were compelled to our attempts to perform our experiments during the day , and the observations therefore , to be taken between 10 p.m. and 2 .
Even during these hours , regularity was far from satisfactory .
As has been explained in the introduction a special room was built for the ments of 1892 .
Here the supply pipe to the motor led directly from the main , the irregularities of pressure were still so great as to render some THE MECHANICAL EQUTVALENT OF HEAT .
imperative .
The form of regulator devised , although it did not insure that absoluto uniformity which we could have wished , was yet so effective that we venture to describe it at length , in the hope that it may prove useful to others who have met with the same difficulty .
AB ( see Plate 4 , fig. 2 ) was a flexible diaphragm made of india-rubber of one ply .
This was bolted between two brass saucer-shaped discs with tinned surfaces .
The lower of these discs was connected with a tube leading to the motor ; the upper , by means of , with an iron reservoir ( a mercury bottle ) placed on the floor .
thin metal rod was attached to a plate in the centre of the diaphragm , and at its lower extremity were fixed two cylinders placed so as to form a balance valve .
The ends of these cylinders were grooved in such a manner that any upward movement dimimished , and any downward movement increased the size of the channels through which the water had to pass .
The water from the main entered at , and afte passing through the grooved channels followed the path indicated by the arrows , until it arrived at the injector of the motor .
An iron tube about 6 feet high was supported by a vertical plank , which reached from the floor to the roof of the laboratory .
The lower end of this tube entered the iron reservoir at the bottom , and to the upper extremity was fastened about 4 feet of high pressure tubin of which the free end was lashed to a mercury jar , similar to the hand one used in a GEISSLER'S pump .
This jar rested on a movable shelf which , by means of a rope and pulley , could be raised from about 4 feet to 10 feet above the vessel .
Sufficient mercury was placed in this jar to fill the flexible tube and the lower 3 or 4 inches of the iron vessel .
The taps at , and , were then opened and water passed into the tube , and the vessel .
As the pressure increased , the mercury was forced up the tube until its surface became visible in .
The tap at was then closed and the screw plug at opened to allow any air in the pipes , or above the diaphragm , to escape .
was then closed and first the tap was opened , then the tap , until the mercury again appeared in H. was then finally closed and would not again be used unless it was necessary to refill the apparatus .
The pressure above could now be adjusted at will by altering the elevation of , and the effective pressure could thus be changed from .
to .
per sq .
inch .
When the tap was opened , if the pressure due to the tap-water on the lower side of exceeded the pressure on the top , the valve was lifted and the flow to the motor diminished until the pressure of the water which had passed the valves became the same as that above the diaphragm .
Conversely , if the water pressure diminished , the flow to the motor increased .
The tap at was only used to remove the pressure of the mercury column from the top of the diaphragm when the instrument was not in use .
Within certain limits , this regulator acted excellently ; but if the water pressure fell so low that the diaphragm assumed its lowest position , it ceased to be effective .
By altering the elevation of the movable shelf at , the rate of the motor could be adjusted at will , and a scale being fixed to the vertical plank , the extent of the adjustment to a given alteration was approximately known after a tions .
[ Note by E. H. G. Added Apri ] , 1893 .
Lord RAYLEIGH has discussed the ciples which should be observed when using a motor of the fan type ( ' Proc. Soc. 1881 , p. 109 ) , and has shown that when the motor is run at high speed the curves work and of resistance cut each other at a greater angle than is possible when rate of revolution is small , and thus a more uniform motion is obtained in the than in the latter , case .
Our speed wheels were so arranged that the motor ordinary circumstances ran at about 16 revolutions per second , which appeared to as high a speed as was necessary .
] SECTION PRESSURE 1N THE SUmROUNDING SPACE .
As already mentioned , the tubes and ( see Plate 3 ) led to the pumps and McLeod gauge .
The circuit embraced several yards of glass tubing , a large number of junctions and four taps , of which three had to be absolutely air-tight .
In addition there were the four insulated junctions in the steel and the two in the calori .
meter lid , and also the six openings ( three in the steel and three in the calorimeter ) , which the glass tubes which supported the calorimeter passed .
All who have worked with low pressures will appreciate the difficulty of making such a series of joints sufficiently air-tight .
The manner in which the difficulty was overcome in the openings through the steel and calorimeter is described in Section VII .
In what we may term the external circuit the test difficulty was experienced in the joints between the glass and the steel tubes and .
Throughout 1891 we used telescopic joints .
The glass tubes , which closely fitted into the steel ones , were coated with hard shellac , and , when both glass and steel had been warmed , were pressed into position .
There is no doubt that such joints can be made absolutely air-tight ; but they cannot be trusted to for any length of time .
It is possible that shellac is sufficiently viscous to long continued pressure , or , on the other hand , it may crack when exposed to changes of temperature .
Whatever the cause may be , we have found such most treacherous .
Their very perfection when first made renders them the for they inspire the observer with a false confidence .
Our experiences during 1891 led us to reject the use of shellac in any form .
* 1892 the connection between the glass and steel was made by platinum tubes small bore , soldered into the steel at one end and fused into the glass at the other .
However , the latter form of junction could not be relied on unless the diameter the platinum was small , and this constriction of both the passages leading to pumps greatly increased the labour of exhausting , for the process of diffusion through * From the ncemenb we avoided the use of india-rubber tubing .
MECHANICAL EQUIVALENT OF HEAT .
such narrow tubes is very slow when the difference of pressure at the ends falls below 1 millim. As we could find no way out of this difficulty we had to content ourselves with a higher pressure than we had proposed to work with .
he taps used were of a kind that has a hole through the stopper a mercury cup at each end ; they app.ear to be very perfect , and any leakage from either end can be detected by its effect on the mercury .
One of these taps was in each glass tube leading from and , and , by their means , .
all the rest of the circuit could be cut off from the calorimeter .
The tube connected with led to a five-way Sprengel and a McLeod gauge ( both by HICKS ) .
The tube from , after passing the first tap , branched into -two arms , one of which led to a Geissler pump , while in the other was a three-way tap establishing connection with the water pump , or with a series of drying bottles , containing .
A tube containing was fixed between the drying bottles and the calorimeter , as we thought it just possible that , otherwise , sufficie fit vapour might pass into the chamber to produce some action on the steel surfaces .
When the apparatus was fixed together the temperature of the tank was raised to about C. ; the air was then exhausted down to about 16 millims. by the water pump , dry air was passed in slowly , and this process was repeated several times .
Finally the pressure was reduced as far as possible by the water pump ( to about 12 to 14 millims and the Sprengel and Geissler brought into play .
It took about 100 double strokes with these pumps to bring the pressure down to 1 millim. , so the labour involved was great , and want of time alone prevented our reducing the pressure further than we did .
By increasing the number of the platinum tubes we hope to be able in future to work with much better vacua .
The absence of leakage is indicated by the fact that the lapse of a week made no perceptible difference in the reading of the McLeod gauge , although a of 01 millim. could have been easily detected .
It was long before this degree of perfection was attained , for the difficulty of ascertaining the position of a slight leakage , in such a circuit , was very great , and many weary hours were expended in leak hunting and in consequent taking to pieces and putting together again the various portions of the apparatus .
SECTION VI .
\mdash ; THE METHOD 0F MAINTAINING A CONSTANT E.M.F. General plan of the electrical Leads passing through points numbered 1 and 3 belong to the Clark cell circuit .
Leads passing through points numbered 2 and 4 belong to the storage cell circuit .
The leads 1 and 2 , as also 3 and 4 , unite at the lid of the calorimeter itself , as shown in the small sketch .
2 The actual position of the galvanometers , \amp ; c. , was different from the sketch , which is only ended to indicate the nature of the arrnngement .
All connecting wires of any length were carried overhead , and hung by silk from a net suspended about 7 bove the table .
Thus all were visible , and could be traced without difficulty .
All keys and instruments through which communication with earth could be made were placed on slabs of paraffin or ebonite , and the accumu .
lators were also insulated .
Each lead was placed as close as possible to its return to avoid any electro-magnetic action .
The resistance of the galvanometer was about 9000 ohms .
Instruments of such high resistance speedily indicate any leakage in the circuit , and it was not until we had given considerable time and attention to the insulation that we were free from trouble on this account .
The galvanometer ( an astatic one manufactured by the Scientific Instrument Company , and the field so adjusted that the time of a single oscillation was about 4 seconds ) showed no deflection unless connected with the Clark cells , although , at times , the difference of potential between the screws 2 and 4 amounted to 10 volt$ If the fingers were placed on the ends of the wires communicating with it , the spot of light could be thrown nearly off the scale .
The galvanometer was placed at the farther end of a tunnel ( 4 feet in length ) of which the nearer end was closed by a sheet of ground glass .
The beam of light was thrown in through the side of tunnel , reflected on to the galvanometer mirror and thence on to the ground glass , which was protected by a hood from other illumination .
Thus the spot\ldquo ; was easily visible , THE MECHANICAL EQUIVALENT OF HEAT .
even in bright day-light , by either of the observers when in the situations they occupied during the experiments .
The extreme sensitiveness of this galvanometer will be illustrated when dealing with the comparison of our Clark cells .
The method of maintaining the ends of the coil at a constant potential difference has been indicated in Section 1 .
The experimental accuracy of the method depends entirely upon the action of the rheostat at .
This instrument should be capable of a fine , as well as a rough , adjustment , and , above all , there should be no possibility of uncertain contact .
The majority of rheostats establish connection by the pressure of a sliding piece upon a wire , which is , at the best , an unreliable form of contact .
The instrument devised by us has , however , proved to be so free from this defect that we describe it at some length .
It is well known that the electrical contact between a moving platinum wire and mercury is , under ordinary circumstances , an unsatisfactory one ; its uncertain nature being probably due to the air-film attached to We found that in vacuo the contact was in every way satisfactory .
and ( see Plate 2 , fig. 3 ) are two glass tubes about 7 feet long , with a platinum wire running through the interior of each\mdash ; the ends of the wires projecting at and The resistance of the wire in is about 6 ohms per foot , and of that in ohm per foot .
and are two wide tubes about feet long , closed at the lower ends .
and were carefully filled through a capillary tube with hot mercury , and inverted so as to stand in and .
Practically , and are two barometers having tubes of about twice the usual length .
and are firmly fixed in a stand , and and are raised or lowered by handles at the side , connected with gearing wheels which move two racks , fixed upon the rods carrying and .
In their lowest position the barometer column reaches almost to the top of the tubes ; in their highest , about 4 feet of the wire is left above the mercury .
Spiral wires are attached to the projectin ends of the platinum wire at the top , and the ends of a copper fork , carrying a screw connector ( E ) at the bend , communicate with the mercury in and The current enters at , passes down the wires in and ( which are thus in parallel arc ) , and leaves the instrument at .
The wire from to is so arranged that additional resistance can be placed in it if required .
The mean resistance of the calorimeter coil used by us this year was about ; the minimum external resistance about .
The E.M.F. of each storage cell was about 14 times the E.M.F. of one Clark cell .
Denoting our external resistance by and the coil resistance by , it is evident that ( when the number of storage and Clark cells were equal ) an approximate balance would be obtained if which gives , i.e. , when the rheostat resistance was about * If the wire is amalgamated the mercury is apt to cling to it when the wire is withdrawn .
It was found that by raising about 6 inches , about 24 inches , and introducin a resistance of into , an approximate balance was obtained .
The resistance was then about , and thus acted as a shunt , and a considerable alteration in it caused only a small change in .
By altering their relative positions the ratio could be changed , and the sensitiveness of the instrument could thus be altered when required .
When arranged as above , a movement of 1 inch in would throw the galvanometer spot off the scale , while the effect of a change of 1/ 25 inch was distinctly visible .
Thus a rise of 1 inch in would change from , i.e. , a change of And thus the ratio would change from to , a change of less than 1/ 50,000 .
The scale on the galvanometer screen had divisions of about 1 inch in length , and a deviation of 1 scale division , during a whole experiment , was exceptional .
It thus appears that the variations in were certainly within 1/ 100,000 of the mean value during the experiments , and , therefore , changes in , consequent on changes in or might be .
It was necessary that throughout each experiment one observer should give his entire attention to the adjustment of the rheostat .
As the temperature of the coil rose steadily the value of had to be increased , its total increase ( over our range ) being about , i.e. , a rise of about 36 inches in B. No doubt this ad.justment took place in steps .
The spot was observed to cross the zero line and was then thrown back to the further side\mdash ; a rise of 1 millim. in being , as a rule , sufficient for this purpose .
As the handle had a large radius , this small change was effected without difficulty .
The effect of such small oscillations about the zero position would probably " " mean out and if not , would be negligible .
Other possible causes of alterations in the value of are\mdash ; ( 1 ) Polarization , as a consequence of the current required to affect the galvanometer .
The high resistance of the galvanometer rendered any perceptible polarization extremely improbable .
In our earlier experiments an additional resistance of was placed in the circuit ; contact was intermittent and established by a key which was worked by a water motor .
Throughout our 1891 experiments we made it a rule to compare before and immediately after each experiment the E.M.F. of the cells used with others which were kept as standards , and , as we were unable to detect any change , these excessive precautions were abandoned .
On twenty-six occasions when no extra resistance was introduced these cells were compared before , and at the close of , an experiment * When the external circuit became heated by the current at the commencement of an experiment oonsiderable adjustment was required , but it was made without difficulty .
This increase is in addition to the increase in due to the heating of the external circuit .
The Olark cells used in nged to the Cavendish Laboratory , and are the ones referred to in Table IX .
, ' Phi ] .
Trans , 1892 , p. 605 .
THE MECHANICAL EQUIVALENT OF HEAT .
POGGENDORFF'S method was used ; a change of one plug indicated a change of each cell was balanced against the standard ones in the manner adopted by Messrs. GLAZEBROOK and SKINNER , and in no case could any alteration whatever be detected .
This point having been satisfactorily established , we considered it unnecessary to repeat the observations in a similar manner this year , but contented ourselves with occasional comparison of the cells used with others which we kept as standards .
( 2 ) The second difficulty was not surmounted so satisfactorily .
As before mentioned ( Introduction ) the laboratory was maintained in cold weather at a constant temperature of about C. , by an automatic lator .
Howeyer , during the of August and September , 1892 , while the ights were cool , the temperature during the day was extremely high , and occasionally rose in the laboratory to over C. The Clark cells were contained in a tank of water , but their temperature at times changed by or C. ; and under such circumstances the difficulty was to know what temperature to take .
Mesf ; rs .
GLAZEBROOK and SKINNER have shown that the E.M.F. does not change with the thermometer , but ] considerably .
A difference between the " " virtual\ldquo ; and " " real\ldquo ; temperature of 1o C. would cause a difference of 1 in 1000 in our corrected times , and we believe that many of the in our results , especially in experiments before August 15 , are attributable to this cause .
On that date we got into working order a regulator which much diminislJed this source error .
] arge bulb of mercury was immersed in the tank , and was connected by a narrow tube with a trap of the ordinary gas regulator pattern , through which cold tap-water , instead of , was kept flowing .
When the mercury expanded , the water ceased to flow through the regulator , and diverged by a bye-way into the tank ; when the mercury contracted , the cold water passed into a waste pipe , instead of into the tank .
The pressure of the water was kept constant by a small supply cistern , with an overilow pipe , which was suspended by a string , and the head of water could thus be regulated .
This apparatus worked most satisfactorily\mdash ; a change of C. turning the inflowing water from the tank to the waste .
Unfortunately the weather became so warm that even the tap-water at that end of the rose to about C. The changes , after the introduction of the regulator , were , however , much diminished , and rarely anlounted to C. As regards the errors thus introduced , although they seriously individual experiments : they must tend to eliminate when a large number of observations are taken .
The Clark cells used for the 1892 experiments were COIJstructed during anuary and February of that year .
Particulars of their construction are iven in an Appendix to Messrs. GLAZEBROOK and SKINNER'S paper ; therefore it is unnecessary to give here more than a brief summary .
Although the method of preparing the for fferent batches of six was varied , the E.M.F. of all the cells is practically the same .
In all cases , the platinum wires were amalgamated by being dipped in mercury when red hot .
The mercury , which was freshly tilled , was about1 centim. in depth .
and was nearly touched the of the zinc rods .
About 1 centim. of the paste\ldquo ; was placed on the , then , about centim. of solution and some crystals of Above this solution was fixed a thin cork , on which was a thin layer of paraffin wax .
On this was placed about 6 centims. of glass-wool and then about centims. of marins glue .
The diameter of the cell was about centims. platinum wire was sealed into a thin glass tube , the end of which projected above the marine glue .
In most casef ; the upper end of the zincs was completely buried .
Nos. to unwashed but shaken up with Hg and solution filtered at zincs not amalgamated .
to Prepared by following out the Boa , rd of Trade directions exactly .
Nos. 49 to 54.\mdash ; Boiled the solution with , allowed to cool to C. Filtered at C. Added to this a small quantity of paste .
Filtered off resulting black precipitate , warmed filtrate , and again filtered at C. The cells made with this solution appeared to settle down to their final E.M.F. immediately after their manufacture , unlike the others , which were some time before they became steady .
About thirty comparisons of these cells were made between ) 1892 .
to 42 remained at the Cavendish Laboratory for some time , and were repeatedly compared with the Rayleigh Standard by Mr. SKINNER , to whom owe our thanks for the trouble he took in this matter .
We select as examples comparisons made on different dates .
The results of the comparisons on other dates are as close , but it unnecessary to give them at No discrepancy showed itself except after rapid changes of temperature , and in the most extreme case the difference from the mean did not amount to 1 in 2000 .
The comparison of cells 31 to 36 with the Cavendish Standard , on February and 20th , gave precisely similar results .
These comparisons were only the last of a series .
On both dates the temperature of the standard was slightly higher than that of the cells , and therefore the equalit was very close .
Mr. SKINNER , writing on February , said , " " You may take it that to 36 are now equal to our standard , and no further comparison is necessary In our comparisons the cells were , in each case , placed in opposition to No. *Tbese are numbered 1:31 to 160 in Messrs. GLAZEBROOK and paper , ' Phil. TranR , 1892 , pp. 622 to 624 .
Numbers 154 to160 were not used by us during this investigation .
Chemicals ) liod by Messrs. THE MECHANICAL EQUIVALENT On May 1 and July 3 the total resistance of the two boxes used was so the numbers given represent the differences in 6000 parts .
In the comparison on September 6 a CLARK'S potentiometer ( constructed by Messrs. ELLIOTT ) was used .
Readings can be taken with an instrument of this kind , with much greater ease and certainty than when using two resistance boxes .
It may be worth mentioning that on Septembel 6 , the instrument was adjusted by one observer and the readings taken by the other , who then moved the handle , thereby a fresh adjustment by the first observer .
The galvanometer used was the high resistance one , , previously referred to , and gave a distinct defiection for a change of 1/ 1 in R. Several comparisons were also made at different dates with some cells belonging * We have not reduced the potentiometer readings bo the same form as that adopted in the previous columns , as a difference of to 1 in 6000 , and the differences shown on September 6 would only amount to ths most case .
to the Cavendish Laboratory .
* These cells were ( early in August ) in close agreement with No. 31 , and also with the ] .
They have since been compared with the Berlin Standard , appear to exceed that standard by about volt .
It will be noticed that the results of the comparison on September 6th are in much closer agreement than those of previous dates .
This is partially due to the potentiometer , ) the adjustment in this case has not to proceed by steps .
It must be remembered also the cells had been kept ( by the regulator above described ) a stcady temperature for a considerable time ; and the great importance of a constant ture is rendered evident by the figures given in Table I. , and i]lustrates the truth of .
GLAZEBROOK'S remark that many discrepa cies are due to the " " " " of the cells when their temperature is changing .
A consideration of the above facfs will , we believe , bear out the conclusion that the mean vadue of our cells may be reg.arded as exceeding that of the standard by not more than volt ( i.e. , by about 1 in 7000 ) .
Duriug our experiments we always used a considerable number of these cells , placing several in parallel and in no did we use less than nine of them at any time .
Thus no correction individual cells is necessary .
A reference to GLAZEBROOK and SKINNEP'S paper will show that but a short time elapsed between their determination of absolute value of the Cavendish standard , and the comparison of our cells with standard .
The absolute value of the standard is given by them as volt ; we are therefore stified in assuming the mean value of No. 31 to 54 , as [ Note by E. H. , added April , .
All the mentioned above practically maintain their relative positions at this date .
I have recently taken a series of observations in the hope of .
some light on the gradual approximation of the cells , as shown by Table I. I find that if the temperature is suddenly raised , and then kept quite constant , the initial differences steadily decrease with lapse of time , and that the cells continue to converge after the lapse of many days , and even weeks .
The cells prepared by my method of adding some to the solution before appear to settle into their state more rapidly than the others .
] SECTION VII .
\mdash ; THE MEASUREMENT OF RESISTANCE .
The coil used by us in 1888 was of iridium vire , and had a mean of about 19 *We take this opporbu1lity of thanking Professor THOMSON for his kindness in permitting us loorrov picccs of apparatus belonging to the Cavendish are the cells Nos. 65 , 69 , 70 , 77 , 78 , and 79 , mentioned in Messrs. and paper .
No. may be regarded as equal to ou No. 31 .
See ' Phil. , 1892 , , 623 , and TrallS , 1892 , THE MECHANICAL OF HEAT .
In 1889 we used two coils of 2 and 4 one of German silver and } of platinum , both covered with a layer of gutta-percha .
In 1891 we used a coil of about 10 covered with amber varnish and made of copper-manganese-nickel alloy , and also a platinum wire of about the same value .
This year we have contented ourselves with a platinum wire of about Since and is , hout each experiment , a constant , therefore Hence if a coil of large resistance is used , the quantity of heat deyeloped will be small , unless is great .
It appeared to us desirable to keep the value of as low as possible , for the greater the difference of potential at the ends of the coil , the greater is the loss arising from imperfect insulation , \amp ; c. On the other hand , if the coil is too short the development of heat is too localized , and its even distribution throughout the calorimeter rendered more difficult .
Our experience in the earlier experiments , as also a consideration of the masses of water , \amp ; c. , led us to the cotlclusion that a difference of potential equal to that of from 3 to 5 Clark cell and a resistance of 8 to 10 would balance the opposing influences in the manner best adapted for our purpose .
Platinum would at first sight appear to be an unsuitable metal of which to form the coil , for its temperature coefficient is very great , and the element of change thus introduced involves much arithmetic in the final calculations .
For the same reason the error due to its rise in temperature the surrounding water is likely to be more serious than if the coil was formed of an alloy .
These considerations carried so much weight that we undertook ( in the summer of 1890 ) a series of tions into the resistance of the various copper-manganese-nickel alloys , which had been recently introduced by Dr. FEUSZNER of Bel.lin .
Some of these alloys were made for us by Messrs. JOHNSON and MATTHEY , and several specimens were given to us by Dr. FEUSZNER , to whom we owe our best thanks for his kindness ; others were obtained from Messrs. WOLFF , Berlin .
The behaviour of these alloys was investigated over a range of from C. to C. The different passed down a tube about 4 feet long , together with a platinum wire , cut from the same coil as that from which one of our thermometers was made .
The resistance of this wire ( in situ ) was determined in ice and steam , and the value of @ being known , by its determination when in our platinum thermometer , the mean temperature of the wires surrounding it could be deduced with great accuracy .
We hoped to proceed with an analysis of the composition of all the wires used but time has not been sufficient .
The results are summarized iu the following table .
Had we certain that the insulation of the wire was perfect , we should have had no hesitation increasing the potential difference , in which case we could have worked largel masses of The uncertainty as to the temperature of the wire when coated would , howc er , more than counterbalance any gaiued by the use of 1 rgor m to September 18 , 1890 .
Platinum wire , diameter000- , same as thermometer ( previously annealed ) .
Copper wire ( commercial ) .
-manganese alluy made by Messrs. JOHNSON and per cent. manganese ( previously annealed in carbon ) .
Copper and manganese ( and M ) , 16 per cent. anese ( previously annealed in carbon ) .
Copper and manganese alloy presented to us by Dr. FEUSZNIBR ( Mn 10 per cent. and Cu 90 per cent Copper-manganese-nickel alloy from Dr. ( diameter 17 millim rather more Mn than and H. Copper-manganese-nickel alloy from Dr. FEUSZNER ( diameter 14 millim Mn 12 per cent. , Ni 3 per cent. , Cu 85 per cent. Copper-manganese-nickel alloy from Dr. FEUSZNHiR ( diameter 10 millim Mn 12 per cent. , Ni 3 per cent. , Cu 85 per cent. Copper-manganese-nickel alloy supplied by WOLFF ( diameter 24 millim Mn , Ni , and Cu , but composition uncertain .
Copper-manganese-nickel alloy supplied by WOLFF ( diameter millim. ) Mn , Ni , and Cu , but composition uncertain .
Copper-manganese-nickel alloy supplied by WOLFF ( diameter millim Mn , Ni , and Cu , but composition uncertain .
German silver .
The above resistances were taken with an E.M.F. of four Leclanche cells , and arms of the bridge were 1000/ 10 .
A double set of electrodes was used , and * Wo ve bo .
C. T. for his in mtlking an analysis of this alloy .
THE MECHANICAL EQUIVALENT OF external resistance determined for each observation .
Also the resistance of tho platinum wire was observed before and after each determination of the other wires .
The current was reversed on each occasion to eliminate thermal effects .
The total swing of the galvanometer about 150 scale divisions for a difference of and thus the next figure could be determined with accuracy .
The actual resistance of the wires used varied from 1 to 50 ; but in the above table the resistance of each at C. is assumed as 1 .
It must be remembered that the temperatures are expressed in the ( not mercury ) scale .
According to the analysis given by ) .
FEUSZNER , and have the same composition , and differ only in their diameters , yet the difference in their behaviour is very marked .
It is possible that this difference is due to the effect of the " " drawing The value of German silver alloy is given for the sake of comparison .
We regret that we did not a wire of that substance .
It is , of course , improbable that its increase is proportional to its temperature\mdash ; from to C. Our observations on these extended from July 24 to September 18 , 1890 , and we were unable to detect any change in their resistances during that time .
The behaviour of the alloy is curious .
Our figures , which were carried to a place beyond that given in the above table , prove that , over the above ran ge the increase in resistance varied strictly as .
In order to test the straightness of its line beyond the above range , we constructed a thermometer which had this wire for its coil , but which was , in all other respects , constructed in the same manner as a platinum thermometer with double electrodes .
This thermometer gave , and , resistance in sulphur vapour at 760 millims. ) .
Assuming the straight line we have According to an investigation subsequently conducted as to the boiling point of sulphur the true air temperature ought to be C. It is difficult to imagine a closer approximation .
A thermometer thus constructed is , therefore , a very simple and accurate instrument .
It would , however , be exceedingly difficult to make another specimen of the alloy of a precisely similar composition , and the above table shows that the effect of small changes may be considerable .
A consideration of the results ] to the conclusion that the wire marked was the one most suitable for our purpose .
Its change in resistance from to C. was only 1 in 5000 , and since the resistance diminished as increased , it was easy to compensate by the addition of a little of the wire .
All difficulties connected with change of resistance consequent on the rise of temperature of the water , or of the All these wires had been previously heated to a temperature of about C. It must be borne in mind that the date of this determination preceded the of CALLENDAR and on this subject , and at this time ( July 27 , 1890 ) the boiling poiul sulphnr was accepted as C. difference in temperature of the wire and the water , appeared , therefore , to have overcome .
This conclusion , however , led to the loss of many weeks ' work , and the cause of much disappointment and delay .
In the summer of 1891 we a series of direct determinations of , the arrangement of the apparatus being nearly the same as that of 1892 .
Our coil consisted of about 30 inches of the wire , , which was first annealed at a temperature of about 20 C. in an of .
The fine white silk covering was allowed to remain , and it was painted Mr. LAURIE'S amber varnish .
* The resistance of this coil , and of the leads from it to the calorimeter lid when situ and immersed in rapidly stirred water , is shown in the following table .
III .
RESISTANCE of the coil at temperature C. , on different dates .
Aug. 22 . . .
23 , 11 . . .
2.5 . . . . .
10.5398 .
, , 27 , before ; after 28 , , , , , ; , , , , \ldquo ; 29 , , , , , ; , , , , 30 , , , , , ; , , , , 31 , , , , , .
, , , , 31 , , , , 10.5953 ; , , . . . . .
10.5977 Sept 1 , . . .
] ; ( some current passed ) The tance of the wire at different temperatures also ascertained in observations taken August 21 to 25 .
It is unnecessary to give the resistance at temperatures other than C. , since resulting curves are all parallel to the one on August 21 .
The alteration caused a rise of C. being only 1 .
in 10,000 , the steady diurnal rise in the resistance the same temperature ) amounts to nearly 1 in 500 in the above interval of time .
trusted that this might ) a result of the annealing , and that the wire would assume stable condition in course of time .
* Mr. LAlJItIE was so kind as to allow us to have a bottle of this varnish prepared by bimself ; for { icnlars this varnish , see ' Proceedings Cambridge Phil. Soc vol. 7 , part 2 , p. 52 .
THE MECHANICAL EQUIVALENT OF HEAT .
It should be noticed that on the above dates we were constantly passing currents through the coil in order to adjust the temperature of the calorimeter during our stirring experiments .
As , however , there was but little current when the resistance of the wire at different temperatures , the curves were not appreciably altered , and we did not recognize the fact that the rise was a consequence of the current .
On August 27 , we performed one of our " " \ldquo ; experiments .
We then had a current of about half ampere passing through the coil for about hours .
We had determined the resistance at the commencement of the experiment , and , on repeating our observations at its close , we found that the value of had risen from to .
It would thus appear that the rise was a function of the current rather than of the time .
We continued the series of experiments until ptember 1 when we decided to abandon the use of this alloy .
* The labour thus expended was noslasht altogether lost , for , as will be explained , in the section on the measurement of temperature , the results of considerable value to us .
Had the rise been a function of the time it would not have been so disastrous , since we could have deduced the .
each experiment .
As , however , the change was caused by the current , it was impossible to determine ( with sufficient accuracy ) the actual value of at each of the experiment .
It appeared probable that electrolysis of some kind took place .
If , after a current had been passing for some time , the ends of the wires from the coil were connected with the galvanometer , a considerable deflection was produced , and it was impossible to determine the resistance of the coil for about half an hour after the completion of an experiment , owing to the disturbance of the galvanometer thus caused .
The effect was apparently due to polarization rather than to thermal , or ' Thomson\ldquo ; effects , and decreased regularly until it might be neglected , the differences of temperature in the circuit were maintained unaltered .
Much time and energy having already been expended in experimenting with these various alloys , we decided to retrace our steps and revert to a platinum wire coil .
Another consideration which influenced us , was that we had , during the past few years , given considerable time and attention to the behaviour of platinum wires when used in platinum thermometers , and we considered that the experience thus gained of the changes in the resistance of such wires under varied conditions , might enable us to proceed with more confidence .
The smaler the diameter of the wire the less its capacity for heat and , in comparison , the greater its cooling surface .
The wire selected for this year 's experiments was supplied by Messrs. JOHNSON and MATTHEY .
It had a diameter in .
, and Mr. SKINNER has recently conducted soms experiments with the same coil immersed in paraffin instead of in water ; and although he has passsd far greater cnrrents than any used by us , he was unable to trace any signs of consequent increase in R. The increase shown during our experiments was , there , probably due to some action between the wire ( when carrying a current ) and the water , this is not unlikely when we remember that one of the metals present was Mn .
NDCCCXCIII .
was of as pure platinum as could be obtained .
The length used was about 13 in .
, and the wire was so disposed that parts near to each other should have only a smali difference of potential .
This wire was covered with a coating of the amber varnish , before referred to .
Several coatings form an excellent insulatin covering for such wire , but we contented ourselves with a thin skin .
The better the insulation the wire and the water the slower the flow of heat across the insulator , and the the difference in temperature between the surfaces which it separates .
If the insulation was not perfecb we have little doubt but that it was sufficient .
In order to throw more light on this poin we decided to make a careful determination of the value of , with the coil immersed in a pure paraffin ( pentane ) , but otherwise under the same conditions as those of the experiment .
We selected pentane ( 1 ) because it would be difficult to find a more perfect non-conductor ; ( 2 ) if pure , it would be easy to free the calorimeter from all traces by passing through it a current of warm , dry air .
latter was an important consideration , for any substance such as ordinary impure paraffin might so alter the interior surfaces of the calorimeter as to change the stirring coefficient .
On December 8 , 1892 , about 250 cub. centims. of pentane*were introduced into the calorimeter , which had previously been dried with extreme care , and the resist ance of the coil was taken at twelve different temperatures .
A summary of the results is given at the end of this section ( see p. 413 ) , and a very slight increase in the resistance is observable , amountin to one part in 22,000 .
It is satisfactory to find that , whatever may be the effect of conduction through the water , it is , in this case , so small as to be unimportant .
The measurements of were made when the coil was immersed and the water stirred in the same manner as during the experiments , hence it follows that if the water acted merely as a conductor it would in no way affect our results .
Even if its conductivity changed with the temperature our values of include the effects of such change , and the manner in which is built up at each temperature is of no consequence .
Again , if electrolysis takes place ( of which , however , we have had no sign ) , polariza tion must ensue .
The best proof of the absence of polarization is the fact that when the storage and Clark cell circuits were reversed ( and this was invariably done more than once during each experiment ) no readjustment of the rheochord was required when the platinum coil , although when using the manganine coil a considerable readjustment was necessazy .
Had electrolysis occurred , it is almost certain that would have been formed .
In our earlier experiments ( both in the platinum and silver calorimeters ) , * Great difficulty was experienced in obtaining this specimen of pentane , and it was only through the kindness of Mr. that we were able to procure it .
It was prepared under his own ection and was free from all impurities , except some of tetrane and hexane .
THE MECHANICAL OF HEAT .
although the same water was nevel used for two experiments , we found that it became acid\mdash ; in some cases strongly so .
This year the water remained unchanged for twenty consecutive experiments and we were unable to detect any acidity , although all water withdrawn from the calorimeter was invariably subjected to a careful examination .
The insulation of the leads passing through the steel cover and the lid of the calorimeter was a troublesome matter , as there was great difficulty in making the insulating junctions absolutely air-tight .
was a brass pin , which had a screw thread on the end and terminated in a nut at A. was a ring of ebonite fitting tightly into the hole in the metal plate and having a projecting flange at the lower end .
The surface of the ebonite was moistened with gutta-percha solution ( made by dissolving gutta-percha in chloroform ) ; the ring , , was then forced into the hole and the brass pin passed through it .
A rin of ebonite , , was then threaded on to the pin and a brass nut , , firmly screwed down\mdash ; the ends of the wire being soldered on at and This form of joint proved to be an excellent one , and when once adopted ( after the failure of many other contrivances ) gave no further trouble .
The steel lid was pierced by such junctions in four places , and the calorimeter lid in two .
As tap-water was continually flowin over the top of the steel lid it was necessary to insulate the exposed parts at .
A piece of wide india.rubber tubing enveloped the wire , the lower end fitted over the ebonite ring , and was wired firmly on to it .
The upper end of this tubing projected above the surface of the water in the tank .
The tubes themselves were filled with distilled water to prevent any rise in temperature of the leads .
The general arrangement is shown in Plate 2 , fig. 2 .
Leads 1 and 3 ( fig. 4 , p. 396 ) were connected with the Clark cells , and thus no current passed along them .
2 and 4 carried the current .
Any heat developed in the parts above passed into the water of the outside tank , and might be disregarded .
portion of the heat developed in the wires and would pass to the water in the calorimeter .
By sufficiently diminishing the resistance of and the heat thus developed could be diminished indefinitely ; on the other hand , the conduction of between the calorimeter and the steel chamber would be increased .
The wire was of pure copper ( No. 21 , B.W.G. ) , and each arm was about 10 centims. in The resistance of was carefully determined , and found to be probable that half the heat generated in these wires would find its way to the and half to the outer , vessel .
We may , therefore , in order to determine the thus added to the calorimeter , assume this external resistance ( r ) as Let be the coil resistance .
Then in any time , the quantity of heat developed in the calorimeter , will be J. T. Now , and since and are maintained at a constant difference of potential , therefore .
therefore the effective value of the resistance .
( neglecting terms involving and higher powers of ) .
Since the mean value of is about , the correction is , in any case , a small one .
In order to tcst the insulation an arm of the bridge was constructed , which included wire No. 4 , the insulators , and the steel casings\mdash ; good connection with the steel being insured by passing a wire into the mercury contained in the thermometer-hole .
High battery power was used , and the resistance was found to be greater than we could measure\mdash ; that is , greater than 10 megohms .
Since No. 4 was in connection with all thejunctions in the plate , the insulation of all the junctions through the was thus established .
The arm was then made to include No. 4 , the insulators in the lid of the calori meter , and a copper wire , which was passed down the gIass tubes until in contact with floor of calorimeter .
The resistance of the wo calorimeter was , in the same way , found to be.greater than 10 megoh1ns .
THE MECHANICAL EQUIVALENT OF HEAT .
tIethod of fmding It .
and ( see fig. 2 , p. 382 ) were blocks of paraffin .
The wires leading from the calorimeter to the junctions on the rod passed through the 1olock , ( lownl which holes were then bored until the wires were exposed .
These exposed portions were carefully amalgamated and the holes filled with mercury .
Two thick pieces of copper were so bent that communication could be established between any of the holes in and the two troughs .
When placed in position , the ends of these connecting strips rested on the conductors at the bottom of the holes .
Two thick wires led from the troughs in to the resistance box , and , in ordel to bring esistauce up to a convenient quantity , a coil was introduced into one of these leads .
Denoting the total resistance of the conductors from the box to the lid of the ca.lorimeter by , it is obvious that all the following resistances could be taken by a movement of the connectors between and .
The determinations were always made in the order thus given .
If , and are the resulting numbers , we have The galvanometer used had a resistance of about , and was fitted with a microscope containing a micrometer eyepiece by ZEISS .
The image of a black line placed on a slip of paper in front of the mirror was thrown on to the micrometer scale .
A galvanometer fitted in this wa is a very convenient instrument , for the swings can , with practice , be read with greater ease and accuracy than when the ordinary lamp and screen arrangement are used .
When a resistance was taken ( whether of the coil or of a platinum thermometer ) the order of the observations was as follows:\mdash ; Having determined between which plugs the resistance lay , the swing to the lower one was taken , the battery inverted and the swing in the opposite direction read .
This process was repeated when the higher plug was in place .
In this way all thermal or other effects are practically eliminated , and the method is independent of small changes in the zero position of the spot .
In some cases , as when working with platinum thermometers at high temperatures , the thermal effects are considerable , and the position of the zero point becomes so changed that the spot is thrown off the scale when connection with the galvanometel is made , although the battery circuit is not connected .
To meet this difficulty a special key was constructed .
The ends of and ( fig. 5 ) and also of and are connected with the galvanometer circuit , the ends of and with the battery .
The screw at is so arrangerl that , when the key is untouched , the galvanomete circuit from the box is completed .
The position of the spot at any time , therefore , is that due to the influence of any E.M.F. in the circuit extel.ior to the battery circuit .
Thus the spot can be placed a convenient position by the colltrol magnet , although a perceptible current is through the galvanometer .
When is sharply pressed down the connection at first broken ; then the battery circuit is established by and and the galvanomete circuit by and ( the projecting blocks on and being insulators ) .
By prope adjustment of the key , the whole operation , performed by one movement , is accom .
plished in a small fraction of time .
Thus the observed swing is due to the differenc of potential caused by the battery and bridge , and is independent of any electro .
thermal effects .
We have described this form of key at some length because , simple as it is , we have not seen any similar arrangement , and we have found it a great conyenience , especially when working with platinum thermometers .
As an example of our method of observation , we give one of our determinations of the value of R. * It should be remembered that , although the exterior leads were made as equal as possible , the measurement of does not depend on their equality and and and are not necessarily equal .
Arms of bridge 1000/ 10 .
E.M.F. that of 2 Leclanche cells .
When measuring and a resistance of was thrown into battery circuit , in order to make the total swing a convenient size .
summary of our determinations of is given at the end of this section see Table XVI .
, p. 413 ) ; the value of in the last column but one of Table IV .
has yet to be corrected for box errors , \amp ; c. THE MECHANICAL EQUIVALENT OF HEAT .
A large number of similar observations were taken at temperatures ranging from to C. Observations at certain temperatures were repeated at regular intervals , and no change of any kind could be detected .
The resulting resistance curve was the characteristic platinum one , and in no way differed from those given by our platinum thermometers .
This in itself seems to us to be an adequate proof that the insulation of the coil was sufficient .
On September 4 , the calorimeter was taken to pieces , and the method of fastening the wire to the glass rods within the calorimeter was altered .
The wire was a little strained during the operation ; but the only effect was to slightly increase its resistance at all temperatures .
A new resistance curve , however , had to be drawn , and new values of used after that date .
It may be mentioned that this coil was often used by us as a platinum thermometer to determine the temperature of the calorimeter when below , or above , the range of The reversing keys in the storage and Clark cell circuit ( see fig. 2 , p. 332 ) were of the ordinary ' rocker\ldquo ; form , but so constructed that as the wires on one side out of the mercury cups , the wires on the other almost immediately made See Section X. The mercury thermometer was used throughout all observations on the temperature of the calorimeter .
contact ; the interval of between the break and make being less than we measure .
If a current be passed through a naked wire , immersed in water , it is certain the temperature of the wire will be greater than that of the water , even when latter is rapidly stirred .
Had we been able to use the alloy wire previously referred to this would have been alnatte of small importance .
Since , however , the value of , when a platinum wire is , varies greatly with the temperature of the wire , we felt it necessary to com. mence a series of investigations into the extent of the change .
Hence we proceeded to determine the resistance of the coil when there was a considerable difference of potential at its ends .
Our first trial was a modification of POGGENDORFF'S method of comparing E.M.F. .
The storage cells were put in circuit through the coil , and a potentiometer arranged in series .
One end of the Clark cell circuit was connected to the storage end of the coil , the other forming a sliding contact on the potentiometer ; thus when the galvanometer in this circuit showed no deflection , when is the resistance of the coil and part of the potentiometer , and the resist .
ance of the rest of the storage circuit .
We had supposed that the internal resistance of the storages might be neglected so that could be determined directly after adjustment by throwing the storages out and completing the circuit by a short length of wire .
The ratio of the E.M.F. could also be obtained by the usual method through the potentiometer , and thus , by noting the temperature of the coil when adjustment was perfect , we had sufficient data from which to calculate the resistance of the coil .
The results obtained by this method were very unsatisfactory , owing probably to the following causes:\mdash ; ( 1 .
) The heating up of the circuit in the same manner as the coil itself .
( 2 .
) The internal resistance of the storages , although almost zero when no current is passing , increases when the circuit is established , and hence ( 3 .
) The ratio of the E.M.F. is not the true one .
Of these , the first was probably the greatest source of error , although this circuit was entirely composed of naked ires , which had a large cooling surface .
We therefore relinquished the attempt to find the whole resistance of the coil , and devised a method whereby we could find the increment of the resistance due to increasing electromotive force ; the resistance when volt having been found at all points of our range as described above ( Table IV For this purpose a WHEATSTONE'S bridge was arranged , constructed as follows:\mdash ; The arm AB ( fig. 6 ) was formed by the coil alone .
contained three large coils of uncovered wire , the first of which ( Cavendish Laboratory coil ) was of German and had a resistance of ; the second , also of German silver ( belonging to : THE MECHANICAL EQUIVALENT OF British Association ) , had a resistance of ; and the third , of copper , had a resistance The arms ( about 6 feet each ) were made of uncovered copper wire .
preliminary rough adjustment could be made at and the final adjustment of the whole bridge was made by altering the temperature of the coil in the arm .
This alteration in temperature was produced by the heat developed in the wire itself , and thus the whole calorimeter was under the same conditions as those during our determination of J. The temperature on thermometer was read at the moment when the galvanometer showed no deflection .
The behaviour of the galvanometer indicated that the flow of heat from the coil was not quite steady , but took place in gushes , owin probably to the coating of insulating varnish on the wire ; for when the key was depressed , the mirror did not swin steadily back to its zero , but oscillated about it , and the moment had to be estimated when the mean of these oscillations coincided with the zero .
That this could be done with considerable accuracy is shown by the ffict that the individual observations agreed well amongst themselves .
The general conduct of a series of these experiments was as follows .
We first found the temperature of the coil when the bridge was in adjustment , the E.M.F. at the ends of the coil bein that used when taking a resistance in the ordinary way .
When the calorimeter was cooled down , the E.M.F. was raised and the coil allowed to warm up until a new balance was obtained and the temperature at that moment noted .
As the temperature and consequently the resistance of the arm remains constant , and since the arms are practically equal , their ratio remains unchanged .
The resistance of the arm AB ( the coil ) must also be the same as before , hence the difference in the two observed temperatures gives us the number of degrees that the coil is hotter than the surrounding water ; or we can express the result in terms of increase in resistance , as the resistance when at all temperatures known .
The total mass of metal in this arm of the bridge amounted to several pounds .
NDCCCXCIII .
TABLE V. The following Table gives a series of observations taken on August 22 , 1892:\mdash ; When the readings obtained by 1 Leclanche differ the change being due to changes in the temperature of the bridge ) , intermediate values are deduced from the times .
It seemed absolutely whether the current was on for only a few secondS or .
In the above set , when three storage cells were being used , the current was not broken at all , the calorimeter being cooled down by the ether in defiance of the heat developed in the coil .
a quarter of an hour elapsed between the observations , they were in very good agreement .
To determine the E.M.F. at the ends of the coil , the various batteries used were compared with the Clark cells .
This method was therefore open to the sources of error mentioned on p. 400 , .
( 2 ) and ( 3 ) .
The resistance of the other parts of the bridge were then determined and were as follows : Resistance of whole bridge large coils coil and its ratio arm .
Hence coil , arm 42 .
large coils and ratio arm .
Hence coil , arm 369 .
, wires forming battery circuit The E.M.F. 8 were determined by the ordinary POGGENDORFF'S in term8 * This observation was uncertain ; the bserver at the galvanometer remarking so , before he that the reading of the temperature was different .
p. 419 , infra .
This would show that the increase in resistance of the other arms of the bridge was not appreciablo THE MECHANICAL EQUIVALENT OF HEAT .
of the Clark cell at temperature C. , hence E.M.F. volts ; set of storage cells being used to give the current through the potentiometer .
TABLE VI .
To determine the internal esistance of the batteries , the current from them was sent through the coil and balanced in our usual manner with the mercury rheostat ( see Section VI The temperature being noted , when the galvanometer was steady , the resistance of the whole battery circuit external to the coil , was immediately taken ; then if is the internal resistance of the battery , the resistance of the coil approximately corrected for the rise due to the current , then , ' but this ratio was given by the potentiometer comparison , hence we found , the internal resistance of the storages , , and of the Leclanches Hence .
we have the following results:\mdash ; * It is probable that the values of are too high , for as before pointed out ) these are the valueb These observations will be seen to agree fairly well with the parabola S This method , although the results were concordant , was unsatisfactory in manner in which the difference of potential at the ends of the coil had to be calculate\amp ; We therefore adopted the following arrangement ( fig. 7 ) , which possessed the great advantage of at once giving us the difference of potential at the ends of the coil in terms of our standards .
The Clark cell circuit was connected to the ends 1 and 3 of Fig. 7 .
the coil , and the rheostat introduced into the circuit ; thus , by adjustment of the rheostat the E.M.F. at the ends of the coil could be made equal to that of 1 , 2 , 3 , \amp ; c. , Clark cells .
No difficulty was experienced in working the two galvanometers together , the observer at the low resistance one being entirely unaware whether the high resistance circuit was made or broken .
As it was felt that the whole success of the method depended on the non-heating of the large coils , the two smaller were replaced by one large one of German silver , wound double , so as to expose a very great sulface to the air .
The resistance of this coil was selected so that when placed in series with the Cavendish large coil , *their united was very nearly equal to that of the calorimeter coil ; thus the remaining arms of the bridge were more nearly equal than before , and were made of stout German silver wire , so that each arm was about 5 feet in length and had a resistance .
We believe that these results were quite satisfactory , since ( 1 ) there was no difficulty in the measurement of the mass of metal in the large coils was very great , their cooling surface large ( a thermometer placed on the coils never.showed any tendency to rise ) and their temperature coefficient small ; ( 3 ) the other arms of the when the exterior resistance is infinitely great , hence the value of ( in the equation ) is too low .
*The Cavendish coil was a triple strand one , containing nearly 1400 feet of single silver wire , the other coil had a double strand and contained more than 400 feet of a stouter wure .
MR. E. H. FITHS ON THE VALUE OF TABLE Hence and the above agree within the limits of experimental error with the parabola SR ; where is number of CLARK'S cells .
In this case there can be no doubt about the actual D.P. at bhe ends of the coil .
The results of the previous investigation ( p. 404 ) are of great importance , in so far as confirm the conclusion that , but are of little use for determining the vaJue of In order to complete these experiments variations in the rate of stirring were tried .
It was found , as ] be expected , that the more perfect the stirring , the less the wire became heated ; but , within the limits of our rate of stirring , the change was so slight that no correction was thought necessary .
The observer at the high-resistance galvanometer could , however , always detect minute in the rate of stirring by the irregular behaviour of the spot .
These changes , although thus rendered very evident , had an exceedingly small effect and thus they served as a proof that the oscill tions observed during our also indicated variations too minute to affect our measuremen [ NOTE.\mdash ; As illustrating the importance , as also the accuracy , of the correction rendered necessary by the difference between the temperature of the wire and of the water , we here give a summary of the results of and .
These two experi ments were performed in order to subject the corrections ( both for radiation , \amp ; c. , and for the rence in the temperature of the wire ) to a severe test , the heat developed per second in the wire during being nine times as great as that developed per second throughout comparing the two columns , it must be that a difference of 0010 corresponds to a difference of but 1 in 8600 in R. THE MECHANICAL hQUIVALENT OF We did not see how to apply a more severe test , and we , therefore , considered it unnecessary to carry the investigation further , although we had supposed that a repetition would be necessary .
] of our Resistance Coils with the Standards .
The resistance box was taken to the Cavendish Laboratory on October 31 , 1892 , to be compared with the standard coils in possession of the British Association .
In order to be able to standardize each coil separately , two plugs were made , half ivory and half brass , so that good contact was secured with the outer circle of the dials through the brass , but no contact was possible with the inner circle through the ivory .
The insulation of these plugs was , however , defective , the shunt so formed bein sufficient to bring the 10legal ohms down to 10 B.A. ohms .
Two brass bridges were then made which rested over the outer circles of the dials , one end of each fitted into the plug-holes , but was cut away so as not to The resistance of the coil had been altered bstween these two experiments ( see p. 411 ) .
touch the central circle .
By tightening uP a screw in the other end of the bridge good and regular contact was secured .
To the centre of each bridge a binding screw was fix connection was made by thick wire leads from these to the CareyFoster In the comparison of the units and tens , the Carey-Foster bridge belonging to the .
was used , and the difference between the standard and the box-unit was found in terms of a length of bridge-wire .
In the comparison of the 's and , the bridge described by GLAZEBROOK ( ' B.A. Report , ' 1883 ) was always used .
* The standard coils used , and their values , are given in the following Table:\mdash ; TABLE X. The same thermometer ( HICKS , No. 352,601 ) that had been employed in all our determinations was again used in the standardization , and , since its bore was very nearly uniform , no correction was necessary .
The B.A. thermometer ( K.O. , 75,149 ) was used for the temperatures of the standards ; its correction was at the temperatures of our observations .
The resistance of the connections between the box and bridge was taken in terms of the bridge-wire of the Carey-Foster bridge .
The results of observations on different days agree well , the average departure from the mean being less than The differences could , in most cases , be easily accounted for by small temperature changes in the interior of the coils , which could not be indicated by the thermometer .
The accompanying Table gives the results .
( Although the coils in the 1000 dials were not used during the determinations of the resistance of the calorimeter coil , we have included them in this Table , since many of them were required when ascertaining the platinum temperatures , and also because a complete standardization of this kind is.interesting , as it shows the order of accuracy of boxes of this pattern .
) *In comparing the 100 and 1000 ohms coils , two Leclanche cells were used .
The currents in the battery were 090 and 028 ampere As the terminals were used for testing at the time the box was the resistance of the connections through the dials was then taken .
This is sufficient to ount for the apparently low value of the first unit , as the resistance of the with all the plugs at is 00334 .
In testing the arms of the bridge , the centre end terminals were used in the same manner as when the box itself formed the bridga so that no plug resistance correction had to be applied .
A table was constructed , giving the total difference between the and real value ( in terms of legal ohms ) for every position of the plugs in each dial .
rence we termed the " " plug correction.\ldquo ; * Having made this correction , we then had to correct for any inaccuracy in the ratio of the arms of the bridge , and as all determinations of the calorimeter coil resistance were made with the 1000 ( right ) and 10 ( left ) as the arms , it is here only necessary to give the correction for these coils .
Now .
This we termed the " " correction The resulting values are expressed in ohms , and true ohmsreading in legal ohms ( 1\mdash ; 0024275 ) .
By applying the corrections in the above form it was possible to perform all the operations by means of the slide rule , as it was only necessary to determine the quantity to be added or subtracted in one case to three , and in the other to four figures .
( When working with platinum thermometers the last two corrections are omitted , as it is only necessary that the resistances should be expressed in terms of the same unit , the absolute value of that unit unimportant ) .
Table XII .
gives the values of , when the calorimeter coil was in water , stirred in the same way as during a " " \ldquo ; determination .
Each number in the column was obtained in the manner given in full on p. 899 , and includes the correction for temperature .
The " " plug correction\ldquo ; cannot be deduced from column , because the differ .
ence between the correction on and has to be used in each case .
The plug correction however is , in every case , about the same as that required by the of and as given in Table IV .
* The correction .
the temperature of the box was made before applying the " " plug correction values of the temperature coefficient for each dial had been previously determined by Messrs. and the accuracy of the coefficients has been exposed to severe tests .
As the temperature of the box was controlled by a special regulator , ib never differed greatly C. Phil. Trans spiral slide ] gives results correct to 1 in 20,000 or THE MECHANICAL VALENT OF HEAT .
Between September 4 and 11 the coil was readjusted , and was probably strained in the process .
The value of was slightly increased .
The sixth figure is , of course , unnecessary .
The numbers in the first and last columns were plotted on a very large scale , so that the fifth figure could be read distinctly .
The resulting curves differ in no respect from those given by platinum thermometers .
The difference between the curve reading ( at of the above temperatures ) and the corresponding value in the table in no case exceeds ( i.e. , 1 in 17,000 ) .
The values of after September 4 exceed the values before that date ( when the temperature is the same ) by 0038 .
The values of ( true ohms ) at other temperatures than those in the above tables were obtained from the curves .
* In order to trace the chatlges in the specific heat of water throughout our range , we had decided to find the time of rising at each of the following temperatures , , and C. It would have been , if not impossible , to arrange our actual observations throughout our experiments so that they were exactly grouped about these points .
The valus of each millim. of our thermometer stem varied as the temperature changed , and the above temperatures in no case corresponded with any number .
The values of our variables at temperatures had to be obtained by As mentioned ( p. 394 ) the resistance of the coil when immersed in pentane rtained .
observations were taken in the same manner and with the rate of stirring as those whose results are given in the aboye tables .
TABLE XIV .
The results must be compared with those in Table XIII .
, since the observations were made after the change in the coil .
In order to find what difference , if any , was caused by the change from water to pentane , the values of in water at the above temperatures were taken from the curve plotted from Table XIII .
large scale curves .
No purpose would be served by reproducing any of these curves in this because if drawn on a small scale they would be useless and misleading ; for they would give an oxag gerated idea of ) eaCCUl.acy of our results\mdash ; the discrepancies being imperceptible .
In every case give data by which the curYes could be reproduced , and where any marked departure is shown by an individual observation we mention it .
SECTION VIII.\mdash ; THE MEASTTREMENT or TIME .
As previously mentioned , an electrical clock ( made by Mr. THOMAS ) was used our standard .
The energy required was supplied by two Leclanche cells .
The dulum ( a seconds one ) was very heavy , weighing about .
Its rod was of woo and we were unable to detect that any appreciable change in rate was caused by small changes of temperature to which it was exposed .
The working parts of electrical clock of this kind are so simple that , when once standardized , it is a reliable instrument .
The pendulum was fitted with a rough and a fine adjustmen and comparisons were made at regular intervals with a rated DENT'S chronometer .
Until August 21 , 1892 , the rate of gain was in 2424 ; after that the rate of gain was certainly less than 1 in 25,000 , and such variations as observed may have been due to temperature changes , as they were irregular and , any case , so small as to render corrections unnecessary .
All observations of made before the above date ( August 21 ) were corrected for clock error , before were used in the calculations .
The actual uncorrected observation is , however , one given in all columns headed " " time The chronograph , which was worked by a separate battery , was of a somewhat peculiar construction .
It was contrived so that on " " making\ldquo ; the circuit it with the clock , the tape was at once started , and , on breaking that the tape came to rest .
Thus the observer could , without moving his position , start or stop the chronograph by pressing a key .
* We rarely , however , used this contrivance , because , as our observations lasted over considerable intervals of time , the instrument had to be continuously at work .
The regulator was so adjusted that the tape passed through about of an inch per second \mdash ; the fractions of a second being rapidly read off by means of glass scales .
The labour of numberin the seconds was great , for the instrument was not so contrived that it could make a distinguishing mark every tenth second .
Immediately before the observer were placed two keys connected with the chrono .
graph , one of which made permanent connection with the clock pendulum , the other was a spring key for recording the times of passing certain temperatures , or the time of each 1000 revolutions of the stirrer .
Some preliminary observations indicated that the personal error ( i.e. , the difference between the actual time and those read off from the tape ) amounted to about of a second .
As , however , all times were recorded by the same observer , this constant " " lag\ldquo ; was of no consequence .
* It was found necessary to place the chronograph at a considerable distance from the galvanometer .
The Table XIX .
, given in Section XII .
, shows that one experiment alone lasted more than 10,000 seconds .
THE MECifANICAL EQUIVALENT OF HEA .
T. SECTION IX.\mdash ; THE MEASUREMENT 0F MASS .
ROWLAND , in the paper to which reference has already been made , takes objection to WEBER'S determination of " " on the ground that he used only 250 grms. of water .
Also M. ARNOUX in the discussion which followed paper said , " " Its determination has always been carried out in diminutive apparatus .
in which the losses by radiation attained a high figure .
With large apparatus the loss by radiation becomes negligible since the volume increases as the cube and its radiatin surface as the square only of the linear dimensions .
With appliances capable of transformin into heat a few hundred kilogrammetres per second , we ought to succeed in determining the mechamcal equivalent to three or four ( sic ) places.\ldquo ; * Now as the quantity of water used by us was small , it will be well to point out the considerations by which we were influenced .
Besides the mass of water , which is used in the calorimeter , account must be taken of what is usually termed the " " water equivalent\ldquo ; of the calorimeter .
[ We take this opportunity of remarkin that when we , in subsequent statements , use the word " " equivalent\ldquo ; we refer to the water equivalent of our calorimeter , and do not use the word in the sense in which it occurs in ROWLAND'S paper as meaning " " the mechanical equivalent ; \ldquo ; this we have designated by the letter " " J Now it is the very difficulty of satisfactorily determining this equivalent , rather than the estimation of loss by radiation , which has led experimenters to prefer a large mass of water , when practicable , in order to make any error in the equivalent one of the second order .
Up to the time of writing , only two methods seem to have been generally adopted for determining the equivalent , ( 1 ) the method of mixtures , ( 2 ) the method by calculations from the specific heats and masses of the component parts of the calorimeter .
Of the first of these we need say but little , for JOULE himself obtained values which varied amongst themselves by 12 per cent. We did not , however , dismiss this method without trial , for having applied it , with every conceivable precaution , to one of our calorimeters , we obtained numbers whose differences were so great as to render them useless for the purposes of this investigation .
The second method is attractive on account of its simplicity , but since the specific heats have probably been determined by the application of the first method , the accuracy is not of a superior order .
Of greater weight is the consideration that a calorimeter has no definite limits , it being impossible to foretell to what extent and in what manner the necessary supports , connections , and surrounding envelopes have to be introduced .
The method which we have adopted of making the experiments * ROWLAND'S calorimeter , when full , weighed over 12 kilos .
, JOULE'S about 6 kilos .
See Table , p. 639 , ' JOULE'S Scientific Papers , ' vol. 1 .
We may quote from Prof. EVERETT'S 'Units and Physical Constants , ' ' Specific Heat of Liquids " " We have omitted decimal figures after the fourbh , as even the second figure is different with different observerS To this we may add that the specific heat of solids will bear no closer scrutiny .
themselves give us the value of the equivalent appears , to us , to be the only way in which this difficulty can be successfully overcome , and , as our results will show , the equivalent was determined with the same accuracy as the other quantities .
* There does not appear to be any argument in favour of using a small mass of water when the value of\ldquo ; \ldquo ; is to be determined according to JOULE'S original method by friction of water ; but the complexion of the case is completely changed if the mods of investigation is altered , for in each different method the accuracy of the result will be some function of the variables at our disposal , and , since we cannot apply the calculus , these ought to be so selected from general considerations , as to make that accuracy a maximum .
In the method which have adopted the variables are ths mass of water , the resistance of the coil , the difference of potential at its ends , and , dependent upon these , is the rate of rise in temperature .
Now the most satisfactory method of ascertainin the rate of rise in temperature , consists in timing the transit of the mercury column across the spider-wire of a reading telescope .
This wire , being set along the graduations of the thermometer , several successive transits are recorded at different parts of the range and the mean time is taken at each place .
Although the rate of rise ought to be fairly great , in order to diminish the importance of radiation losses , if the rate is too rapid the observer has not sufficient time to set his cross-wire to the successive graduations , and only one observation can be recorded at each part of the range .
The confusion which would be introduced by attempting to observe every other scale division need scarcely be mentioned .
It was only by constant practice that one of the observers was able to take observations with accuracy at the rate we adopted .
If , on the other hand , the advance of the column is too slow , the time of apparent contact with the telescope wire is uncertain and , at the same time , the phenomenon of " " sticking\ldquo ; is rendered more probable .
As during these experiments the strain on the observer was very great , mere personal fatigue would be sufficient to prevent accurate observations being taken if the duration of the experiment was extended beyond an hour .
In the case of the experiments on the heat developed by the stirrer , it will be seen by reference to the tables that these sometimes lasted three or four hours ; but the nature of these experiments not being so complicated , the strain was less and the observer had short tervals of rest .
The rate of rise in temperature thus practically determines the values which must assign to the other quantities at our disposal , for to make the rate great we must either diminish the mass or resistance , or else increase the difference of potentiaL The considerations which led us to give to and the limiting values assigned to * The value of can be obtained without ascertaining the valu of the water equivalent ( see p. 481 ) , and the determination of the changes in the capacity for heat of water is entirely independent of it .
Our conclusions are based on observations of in time caused by differences in the mass water , and we see no reason why such differences could be ascertained with greater accuracy if the equivalent was diminished .
total times would be decreased , but the diflerences wouJd remain unalteM THE MECHANICAL EQUII* , ALENT OF HEAT .
them during these experiments have already been mentioned on p. 389 .
We were thus restricted to masses of water .
from 130 to 800 grms. As the determination of the equivalent and the in capacity of the whole mass due to increase of specific heat are fully discussed in another section ; it is only necessary here to deal with the mass of water added to the calorimeter .
The balance used by us was by BECKER , carrying grms. and turning with a milligramme .
As this type of balance is not provided with a rider-arm , it was found convenient to add weights to the nearest above and below , and to interpolate by the method of oscillations .
Nearly the whole of the surements were conducted by the ordinary method of direct weighing , as the balance , when once in good adjustment , remained so .
The adjustment was however , occasionally checked by the method of double weighing .
Holes were bored through the bottom of the balance case and the top of the cupboard on which it stood , and wires suspended from the pans for use when weighing flasks , \amp ; c. As it was necessary to open this cupboard whenever a flask had to be weighed , it would have been difficult to keep the air within the cupboard in a state of , and thus if proper precautions had not been taken a considerable error might have been introduced in the weighing of glass vessels of large area , owing to the hygroscopic nature of their surfaces .
The usual simple expedient of taring one glass vessel against another of approximately equal surface appeared eminently satisfactory , as the weighings could be repeated at varying intervals of time ( generally a few hours , but in one test case four days ) and the results were always found to to a ramme .
In our earlier series of experiments ( 1887-1890 ) the calorimeter be apidly taken to pieces , dried , any required mass of water added by means of a fine dropping- pipette , weighed , and put together again .
As , however , those experiments had shown us the necessity of keeping the outer space perfectly dry , we now decided to reduce , if possible , the pressure to less than 1 millim. , since , in this case , radiation , \amp ; c. , would be considerably diminished , and any moisture present would immediately be detected .
It thus became impossible to take the calorimeter to pieces , and means had to be devised for introducing and withdrawing water through the supporting glass tubes .
The impossibility of removing the calorimeter without destroying our vacuum led to another difficulty , for , at the end of the experiment , the water had been heated to the highest point of our range , and before we could conduct a second experiment we had either to cool the whole apparatus down , which would have been an extremely slow process , owing to the smallness of the radiation , and would also have rendered it necessary to throw the regulator of the outer bath entirely out of adjustment ; or else withdraw part of , or the water in the calorimeter , adding sufficient cold water to bring the temperature down below the point at which our observations commenced .
During the earlier part of our 1891 series we withdrew a portion only of the water by means of a modified form of weighed pipette , and then , by the same means , added a weighed quantity of cold water .
This method was in some respects satisfactory , NDCCCXCIII .
that every withdrawal involved four weioo.hings , so that , after a short time , the of water in the calorimeter was dependent upon a very great number of , and small errors in each , if cumulatiye , might have become serious .
therefore , soon abandoued this as a general method , and only resorted to it occasionally .
Since then we have always used the following method of adding a known llass of water to the calorimeter:\mdash ; Pairs of glass flasks ( with india-rubber stoppers ) , of 100 , 200 , and cub. centims. acity , were selected and carefully cleaned .
One of each pair kept as a tare for its fellow .
In introducing the water into the calorimeter it was necessary to avoid loss either by accident or by evaporation , and to guard ainst losses of this kind the following device was employed : thin glass tube , 30 centims. in length , had a wider piece , about 7 centims. long and 2 centims. in diameter , fused to its upper end , thus forming a fullnel .
Up the centre of this wide portion a piece of glass rod was fixed , its upper extremity projecting above the top of the wider tube , so that the lip of flask , from which water was poured , could rest against it , and thus prevent any drops running down the outside of the flask or tube .
By forming a slight support , it also somewhat relieved the difficulty of the operation , which , even with its aid , was a trying one .
A cork was fixed to the stem of the funnel , in such a manner as to permit its lower end to dip into the calorimeter .
This funnel was provided with a case , consisting of a length of glass tube closed permanently at the lower , and temporarily at the upper , end by an bber stopper .
A piece of similar , closed at both ends , was used as a tare .
The full flask and funnel having been weighed , the funnel was withdrawn from its case , placed in one of the tubes leading to the calorimeter , and the water poured down it from the flask .
This is the only part of the process during which evaporation could possibly take place , and any such loss , during the few seconds the operation lasted , would be a matter of no consequence .
The empty flask was immediately re-stoppered , the funnel withdrawn and returned to its case , and the weight ascertained at any convenient time .
This method also enabled us to add water below the dew-point of the room without any error due to moisture condensed on the flask and funnel , for it was only necessary to first weigh the flask , then cool it , and pour its contents into the calorimeter .
The exterior of the funnel was dried before returning it to its case , and the stoppered fiask laced in a dessicator for some time before the final weighing .
This method appeared to be satisfactory , and , although it was impossible to add exactly the same amount on different occasions , still , as the difference was only the fraction of a gramme , it was easy to apply any necessary correction to the resulting numbers .
A glw tube , reaching to the bottom of the calorimeter , was connected with an exhaust pump , which enabled us to withdraw nearly all the water , and the remaining moisture was removed by passing a current of hot dry air through the calorimeter for several hours .
Owing to difficulty with regard to the air-tight joints , we were obliged to take the calorimeter to pieces at the beginning of July , 1892 , and to our astonishment fECHANICAL hQUlVALENT OF we found , although we had , as we thought , dried completely , there still remained several cub. celltims .
of water .
We subsequently ascertained that the calorimeter could be properly dried in the above mannel , provided that sufficient time was allowed ; but the hot air had to pass .
several days before a drying tube , suspended within the calorimeter for 12 hours , showed no increase in weight .
Now , as the first method involvea too many weighings , and the second was unsafe , unless two or three days elapsed between each experiment , we were compelled to seek a different mode of procedure .
Thus , as we could neither withdraw any portion of the water , nor cool the whole calorimeter down without great labour , means had to be devised for cooling it from within .
A tube , through which a freezin mixture could be forced , was evidently objectionable , as there would always be uncertainty as to whether all the freezin mixture had been removed .
We found the following arrangement worked admirably , for the temperature of the calorimeter fell as rapidly when cooling , as it rose when the current was passing .
We were thus able to use the same water )gain and so save a numerical reduction .
A very thin glass tube , closed at its lower end , passed down the support tube ( Plate 2 , fig. 1 ) of the calorimeter , just clearing the base , and the upper end , which was open , projected about inch above the top of the tube , .
Care was taken to so fix this tube that no water could find its way up the annular space between the two ubes .
When not in use , the upper end was closed by a cork , so as to prevent air.currents circulating down to the calorimeter .
When used for cooling , this tube was partially filled with ether , through which a rapid current of dried air was sucked by means of the arrangement shown in Plate 2 , fig. 1 .
The fine capillary poin reached just to the bottom of the closed so that , when the cooling had far enough , the remaining ether could be withdrawn by reversing the connections to the exhaust pump and the drying bottles .
Any ether or impurity left at the bottom of this tube would have caused great irregularities in the experiments , so , as a precaution , only freshly distilled ether was used , and when the temperature had been sufficiently lowered , dry air was passed for a considerable time and , finally , the cooling tube thoroughly cleaned by means of an absorbent mop .
Thejoint at was made by slipping a small ring of india-rubber tubing round the projecting end of the cooling tube , and , by slightly splaying out the mouth of the aspirating portion , a sufficiently good union was effected , the extreme thinness of the walls rendering a ground glass j ) able .
The entrance tube was for adding fresh ether as occasion required .
As soon as we perceived that this method of cooling was a success , it was at once evident that it could be applied to finding the latent heat of evaporation ofliquids , for by regulating the rate of evaporation we could make the cooling effect such as to counterbalance the heat developed by a known current .
We have already made a few rough determinations , but prefer to postpone their publication until we have applied the method with greater precautions .
* The ether was always redistilled by us immediately before use\mdash ; at a temperature under C. SECTION X.\mdash ; THE MEASUREMENT 0F TEMPERATURE .
When commencing this account of our experiments , we proposed to give full of the calibration and standardization of the mercury thermometer , upon which our measurements depend .
Nearly half of our time and attention has been to this ) of the subject , and thsre are but two courses open to us , viz. , to give full account of our method of standardization , with the tables upon which our con .
clusions are based ; or reserve a detailed description for a further communication , and give a short summary of the results in the present paper .
We have decided to adopt the latter course , for the following reasons:\mdash ; ( 1 .
) We are ( with the help of Mr. CALLENDAR ) now entering on a careful direct comparison of thermometer with a new form of air thermometer , which , there every reason to believe , will give very accurate results ; but we are unable to any definite limit to the time that this investigation may take .
( 2 .
) Dr. GUILLAUME is engaged in the comparison with the Paris standard a thermometer specially constructed .
us by M. TONNELOT , and some time must elapse before this comparison can be completed .
( 3 .
) The dimensions of this communication are already very great , and if , in addi .
tion , we gave a full account of our thermometric comparisons , its length would become immoderate .
( 4 .
) We consider that the success of the method of exact determination of temperature by platinum thermometers may now be regarded as established , and that the accuracy of the results obtained by platinum thermometers in our possession is proved by communications already published .
( 5 .
) The results of our experiments may themselves be taken as a proof of the accuracy of the method of time-calibration adopted by us , and described on pp. and 424 , infr Assuming the accuracy of the calibration it is possible we are in error\mdash ; ( 1 .
) In the actual elevation ; ( 2 .
) In the value of our total range .
No. ( 1 ) may be disregarded , for it is only the changes in the specific heat of water which render the actual clevation of any consequence whatever .
Now it is impossible that we are in error by C. , for the comparisons we have made with the B.A. thermometers , which have been carefully standardized at Kew , would alone been sufficient to detect any error of that kind , and an error of C. in would mean an alteration of only 1 in 20,000 in the value of J. No. ( 2 ) is , of course , of vital importance .
In ' Phil. Trans vol. , will be found the results of a comparison of * Dcccmber , 1892 .
'B.A .
' 1890 ; 'Phil .
' vol. ] and 'Phil .
Mag ; 'Chem .
Soc. ' 1890 , THE qCHANICAL EQUIVALENT OF T. 42 ] of our platinum thermometers ( H ) and the air thermometer over a range of to and the numbers there given show that the probable limit of error at point in the above range is C. This limit may be further diminished by the " " plug corrections\ldquo ; ( see p. 410 ) , then neglected by us .
It must be remembered that this limit of error refers to the actual elevation , and is probably decreased when we come to consider differences over a range of C. Further , the table there given is a record of only a single series of observations , and must include experimental errors , which would tend to mean out if sufficient observations were taken .
Again as regards differences , the quantity which chiefly affects us ( apart from experimental errors , etc. ) is the value of S in the equation ; and it is therefore important to point out to what extent the variations we have observed in the value of S may influence our results .
We have standardized our mercury thermometer by means of the platinum thermometer particulars Phil. Trans This thermometer contained two wires , and by placing these in series a third resistance could be determined ; it thus practically contained three coils , which we denote by , and B. The fixed points of this thermometer have been re-determined at regular intervals during the past two years , and no change of any consequence has been observed .
The values of its constants are as Now , the value of at C. and C. obtained by assuming the above values of 8 are and , and it is on the difference between these numbers that possible error in the value of our range depends .
Assume it as possible that the true value of S is as high as 1.70 or as low as 1.60 ( our greatest differences in any determinations of this constant have been 009 ) we should then The coil of thermometer was so long that ib was barely contained in the calorimeter , and , as it required complele immersion , it was not suitable for use in this case .
The constants differ slightly from those previously published ( Phil. Mag December , 1892 ) , as the plug correction\ldquo ; ( see p. 410 ) , has , for the first time , been applied to them .
By assuming that we get for the absolute zero - , and instead of the numbers there given .
and as the differences , thus causing changes of 004 and 008 , respectively , in our range , although causing differences of 010 and 008 in th9 eleyation .
Apart from errors of observation , we may , therefore , regard the limit of our error as in C. , i.e. , 1 in 4000 ; and we believe it to be much less .
As to experimental errors , it must be remembered that we have made th6 comparison of the platinum and mercury thermometers under the most favourable circumstances .
Both were immersed in the calorimeter , the cooling tube ( Plate 2 , fig. 2 ) , having been removed to allow of the insertion of the platinum thermometer .
observer , with his eye at the telescope and his hand on the key could , by means the current , set the temperature with extreme exactitude to any required reading stirring bein maintained at the same rate as during an experiment .
Observations could thus be repeated at the same reading as often as necessary .
Again , the observations taken by the observer at the galvanometer were entirely independent of the readings of the observer at the telescope , yet a rence of C. , on the repetition of an observation , was very unusual .
Many hundreds of comparisons were made , the particulars of which we reserve for our second communication .
We may mention that the comparisons made in 1892 differ from those of 1891 by 003 in range , but show a rise of about in the actual elevation of was one of three thermometers specially constructed for this work by HICKB .
We set him a somewhat difficult task , as the dimensions of the bulb were strictly limited both in length and diameter , and the weight of mercury was ] specified .
The stem was divided into millimetres , about 40 millims. C. , and the lower 18 centims. were out of sight within the calorimeter tube , the graduation marked being within this tube and the 26th millimetre just appearing above it .
The reading C. ; the reading C. The graduations continued to 580 millims. Our method of observation obviated the necessity of attempting to divide the millimetres by eye .
The thermometer was fixed in such a position that the view of the bore was unimpeded by any of the shorter graduations .
When timing changes of temperature the spider-wire of the telescope was placed on these graduations and the time of contact with the spider-wire observed .
No attempt was made to take any reading terminating in or 5 , as these graduations extended across the bore and obscured the edge of the mercury column for about } illim .
Thus to obtain the time of pa , ssing a given point we were limited to the observation of four consecutive transits , and for this reason , our ' fixed points\ldquo ; were determined at , \amp ; c. , and the time of passing these points deduced from the times at 86 , 87 , 88 , and 89 , \amp ; c. A consideration of the behaviour of this thermometer ( as well as of other narrow bore ones ) rendered it evident that the ordinary method of calibration was insufficient MECHANICAL OF HEAT .
if the thermometer was to be used for the determination of changing temperatures .
We believe that no calibration can be considered satisfactory unless made under the same circumstances as those in which the thermometeris to be used .
As an illustration of our meaning we may mention that the range to 1275 C. when the thermometer was stationary , but C. when rising at normal rate of our experiments , i.e. , in addition to calibration for inequalities in the bore , a thermometer used under circumstances similar to ours must be calibrated for " " sticking\ldquo ; and for certain periodic alterations in the rate of advance which are probably due to the behaviour of the glass bulb .
Whatever may be the case with other thermometers , a close and prolonged study of the thermometer has rendered it certain that the rate of advance of the column ( when the rise in temperature is uniform ) is influenced by ot , her conditions than mere irregularities in the bore , but that these changes in rate will , under the same conditions , almost always reCdr at the same places .
During 1891 the times of rising every 10 millims. *throughout the whole range were observed and the observations repeated on more thirty occasions with varying rates of rise .
Nearly half of these observations were taken with he alloy wire , whose resistance did not increase as the temperature rose , the remainder with a platinum wire .
During this series of observations the weights of water were , the supply of heat was trebled , the exterior temperature was shifted from the bottom to the top of our range , the pressure changed from 1 to 760 millims. , and , in fact , all the conditions were altered .
The times of rising 10 millims. under all these different circumstances were thus determined at 42 points in our range , and the results plotted ; the mean path in each case was , of course , different , but the deviations from it were regular in their occurrence and similar in their nature .
Such deviations as invariably occurred throughout all the altered conditions must have had their origin in the thermometer itself .
We hope to publish some of these curves in our further communication , although the changes in the curves themselves are great , the steady repetition of lag and acceleration is both curious and interesting .
This year these differences recurred with another coil and a different method of stirring .
The mean values of these deviations , which were repeated throughout the whole of the experiments , were expressed in terms of the curve-reading at each point and plotted as ordinates on a straight line , and the remainder of the calibration com- pleted in usual manner .
The effect of the application of the resulting corrections to the numbers obtained from our experiments is a sufficient proof of the accuracy of the method .
An error of C. in the comparative value of any of our ranges would produce a difference of from to second between the " " mean times\ldquo ; 10 millims. on thermometer C. approximabely .
E.g. , if was observed time over a small range , and if was the value over that range as given by the smooth curve , then was plotted as ordinate and as abcissa , areas ) obtained were integrated , \amp ; c. in , 9 , 14 , 18 , and 2:2 , Table XL .
, and the curve numbers in the columns .
Further , this difference would recur in experiment .
No discrepancy of any consequence is obseryable .
In our own work we have the value of these ranges to another figure , but we refrain printing it until produce more evidence .
One possible canse of error in our total range yet viz. , the difference caused by unequal lag at the beginning and end of that range , this difficulty cannot be surmounted by means of the same method that used determining the comparative values of the smaller anges .
Our experiments on this point were of the following nature:\mdash ; A platinum thermolneter ( Pt ) of exceedingly small capacity for heat ( the external diameter was only 3/ 16 inch ) was placed in the calorimeter .
The coil of this there .
mometer was pressed tightly against its glass envelope , which was of egg-shell thin ness , and thus the thermometer was rendered extremely sensitive .
Its stem passed down a condenser-tube , which the tank water was continually drawn by means of the water pump .
The stem-resistance , therefore , remained constant , and thus only one observation was required , whereas with three observations had to taken to accurately determine a temperature .
The value of when the temperature of the calorimeter was steady was determined at the initial and final points of our range .
The temperature was then lowered to our usual raised in the same manner as an experiment .
The value of was when the reading of ( mercury thermometer ) again indicated the initial and points .
No doubt the platinum thermometer also lagged , but we proved by inde .
pendent experiments that it did so to an extremely small extent , and , if uniform , the lag was of no consequence , as it would only affect the elevation and not the Now in the platinum thermometer we have no ' sticking \amp ; c. , and there is every probability that the lag regular .
The observer at the galvanometer called at moment when the " " spot\ldquo ; was seen to pass the zero point , the other observer the mercury readings at the same instant .
This was done over four or five tiye plugs , and the resistance deduced at the moment of passing the required The results indicated that the lag of the mercury thermometer was greater at commencement than at the end of our range by ( at our normal rate of rise ) about The " " stationary range " " would , therefore , require diminishing by about in C. , or by 1 in 1400 .
In order to investigate this point more fully another platinum thermometer having a naked wire , was placed in the calorimeter .
The coil of this thermome had fifty turns to the inch , and thus the different parts of the wire were very together .
The results were not satisfactory , the conduction across the water , * The average lag of the mercury thermometer proved to be less than we anticipated .
It , of depends , in a measure , on the thoroughness of the stirring ; but when the mass of mercury bas to bs heated considered , the extreme lag , observed at our rate of rise , , 025 , is small .
THE MECHANICAL EQUIVALENT OF HEAT .
such circumstances , being sufficient to render the galvanometer uncertain and to produce irregularities in our results .
We propose to repeat these " " lag\ldquo ; determinations , using the naked wire , but placing pentane instead of water within the calorimeter .
In the meantime we have left our total range uncorrected for the difference in the imtial and final lag , as our own data are not at present sufficiently accurate for us to apply the correction , which is certainly less than 1 in 1400 .
Its effect would be to increase the value of but it would not affect the changes in the specific heat of water or the calorimeter .
The correction will only have to be applied to the final result .
The exterior temperature was determined by a mercury thermometer ( by HICKS ) , which was similar in most respects to , but its range was not so open\mdash ; about 27 millims. C. It was placed in the calorimeter with and the readings compared at every 5 millims. of .
When conducting our experiments we only required to know accurately the difference between the internal and external temperature , and , for such a purpose , this comparison was sufficient .
As A was to be used for the reading of stationary temperatures , a careful calibration of the usual kind was made by means of the Cavendish dividing microscopes .
Our method of experimenting eliminated several of the errors which are associated with the use of mercury thermometers .
] .
Changes of temperature of stem .
The exposed portion of the stems of A and was always immediately above the surface of the water in our twenty-gallon tank .
The water was always maintained at a constant temperature see and ( as we ascertained by direct measurements ) the temperature of the air above it varied but slightly .
The lower portion of the stem passed through a tube immersed in this tank water , and thus the stem temperature remained steady when the temperature of the bulb was rising .
The change in the temperature of even the upper portion of the stem never exceeded three or four degrees throughout the year , and the change in the average stem temperature must have been much smaller .
At the time of determination of the values of the ranges , the stem was , therefore , under exactly the same conditions as those prevalent during an experiment .
( 2 .
) The effect of changes of pressure on the bulb .
The only changes to which it was exposed were atmospheric ones , which , even if they affected the elevation , would not affect the range .
Again , A and were so ROWLAND appears to have paid no tention to the effect of this difference in initial and final It is probable , tlJerefore , that the correction , if it was possible to obtain it , would cause a greater increase iu his value than it is likely to effect in ours , since his average rate of rise was three times as great .
This calibration is in practical agreement with another calibration of A conducted by platinum thermometers ; the inequalities as determined by the latter method are , however , always greater than those shown by the former .
This is to be expected when the length of the mercury used in the ordinary method of calil)ration is considered .
NDCCCXCIII .
similar in dimensions that the result of such changes would be the same in both .
Thus the differences of temperature deduced from the would not be altered to any appreciable extent , and it was only with such that we had to deal .
( 3 .
) The changes in zero caused by the sudden changes io temperature .
The history of thermometer was much the same from day to day .
During night it remained about the temperature of the tank .
Before an experiment , temperature was lowered to C. and , an experiment , it was steadily from C. to C. , no readings being taken until it reached C. When determining the fixed points , the same routine was , as nearly as possible , observed ; thus the were , in all cases , so alike that the effects were probably In any case , we have been unable to detect any irregularities arising from this TABLE XVII.\mdash ; Values of\ldquo ; Fixed Points\ldquo ; on Thermometer when the is Steady .
The following Table shows the difference between the comparative values of th6 ranges when the thermometer is steady and when rising .
remainod in the calorimeter from the commencement to the cIose of our work except on few occasions when water was added or withdrawn .
THE MECHANICAL EQUIVALENT OF HEAT .
The calibration of those portions of the thermometer between our " " fixed points " " was conducted in the same manner as the calibration over the whole range .
[ Note by E. H. GRIFFITHS , added 1st , 1893 .
On February 14 I received the thermometer by M. TONNELOT ( No. 11048 ) , which had been standardized both by Dr. GUILLAUME , and under his direction , at the Bureau International des Poids et Mesures .
This thermometer is of hard glass , with transpareDt stem , divided into fenths of a degree Centigrade .
The scale is continuous from \mdash ; 3o to C. when ther ' occurs a bulb , then a portion giving readings to C. , then a second bulb , and , finally , readings from to C. The length of each degree is about 10 millims. The study of this thermometer has been a most exhaustiye one ; two separate calibrations were performed for the most useful part to the difference between any two values given by the separate determinations being C. The fundamental interval and the coefficients of external and internal pressure were repeatedly determined .
Several hours must have been occupied in the mere copying of the observations forvrarded to although the record consists , almost entirely , of numerals .
I mention this as indicating the labour which has been devoted to this standardization , and I owe my warmest thanks to the Bureau International for the care and attention bestowed upon it .
The papers include printed tables for the reduction of the corrected mercury readings both to the hydrogen and the nitrogen scale as resulting from the comparisons of Dr. CHAPPUIS .
The comparison between this thermometer ( denoted hereafter by P ) and the thermometer was conducted in the outer tank , for had the comparison been made in the calorimeter , the lower readings of would have been invisible .
was , therefore , under the conditions as during our experiments , except in the following particulars:\mdash ; ( 1 .
) The external pressure was increased by about 110 millims. of water .
A series of observations proved that the resulting correction was millim. ( rather less than C. ) and this correction was always made before entering the reading of ( 2 .
) When in the calorimeter , the lower portion of the stem passed tbrough tubes washed by the tank water .
About 60 millims. were tbus at a constant temperature , whatever the extent of alteration in the bulb temperature .
millims. intervened between the top of the calorimeter and the lower end of this constant temperature portion .
We may consider that this 50 millims. assumed the mean temperature between the tank and the calorimeter .
Thus a length of millims. of the stem , which during our previous experiments was at the tank temperatnre , assumed , this comparison , the temperature of the bulb .3 1 2 MR. E. H. GRIFFITHS ON THE VALUE OF The correction in millimeters is given with sufficient accuracy by the formula and as the mean value of 1 millim. is C. , the resulting correction is This amounts in the greatest case to C. The readings of were taken by a microscope ( re-levelled before each observation ) which was fitted a micrometer scale .
The one hundredth of a degree Centigrade was thus read directly and no difficulty was experiellced in estimating the one thousandth , the scale being an unusually clear one , made by the Cambridge Scientific Instrument Company .
The thermometer was placed in a tube , open at the lower extremity , and an opening at the upper end was connected with a rubber tube leading to the waste .
This arrangement was used as a siphon and the thermometer stem was thus maintained at the bulb temperature , and , therefore , as the mercury column rose , any alteration in the stem immersion was rendered unnecessary .
In order that the readings might be obtained under the exact conditions observed by Dr. GUILLAUME , I implicitly followed the instructions given in a printed note added to the certificate .
After the tank had arrived at the required temperature , at least half-an-hour was allowed to elapse before taking a reading , and immediately after the observation the thermometer was transferred to ice , and a series of readings taken to obtain the greatest zero depression .
The admirable methods adopted by the Bureau International are less known , or , if known , are less practised in this country than they deserve to be .
I therefore give one example in full showing the : corrections .
THE MECHANICAL EQUIVALENT OF HEAT .
COMPARISON of the Reading 875 on with ( March 15 , 1893 ) .
Temperatur reading by P. Barometer ( corrected ) pressure ( expressed in terms of Hg ) Total external pressure Observed reading . . . . .
Calibration correction External pressure correction .
Internal Zero . . . . .
Fundamental error correction . . . . .
Sum corrections Correction for stem see , Hence reading on the mercury scale Correction to ] .
to scale Hence Reading .
Hydrogen scale .
Nitrogen scale .
87 .
13.985 I confess that I did not know until I adopted in detail the precautions advised by Dr. GUILLAUME in his ' Traite de Thermometrie that it was possible to regard the mercury thermometer as an strument of precision , and I had considered such corrections as too refined for so rough an instrument .
I am , however , now convinced that , given the right conditions , it is possible to ascertain temperatures by a mercury thermometer with a limit of error of or .
It is rarely , however , that such conditions prevail .
The labour involved in the previous standardization of the thermometer is enormous , and there are but few men capable of constructing thermometers on which such labour may be profitably expended .
The actual operation of observing and reducing the results is also considerably greater than is the case ] ) platinum thermometers are used .
The following table gives the results of all observations made with the micrometer eye-piece .
I had previously made several compal.isons whose mean result is in fair agreement with those here given ; but , as the last two figures of the readings had to be estimated , the discrepancies are greater , the results of less value .
THE MECHANICAL EQUIVALENT OF HEAT .
It thus appears that our temperatures , as deduced by platinum thermometers , exceed those of the Paris Hydrogen and rogen Standards in actual elevation by C. and C. respectively , but that the value of our total temperature range is practically the same by both methods of standardization .
* It is true that slight discrepancies appear in the value of the smaller ranges , but here I am inclined to place greater reliance on our previous determinations than on those obtained by the comparison with , for mistakes of observation are far less likely to occur when using platinum , than when reading mercury , thermometers , and also the former values are deduced from the observation of several different resistances at each temperature , all of which after the correction given by the standardization of the -box are in practical agreement , although the actual values of differ greatly .
The difference in actual eleyation is of no consequence , although it is interesting to notice that the agreement with the Paris Nitrogen Scale is nearel than that with the Hydrogell Scale .
This was to .
be expected , since the -point of sulphur waq ascertained by Mr. CALLENDAR and myself by means of an tbermolneter , and our values of depend on that determination .
We have not , as yet , been able to commence , by means of Mr. CALLENDAR'S new of air thermometer , our direct comparison with the nitrogen scale , for calibrating and cleaning of the instrument have occupied much time .
Pending the completion of that work the results of the above comparison are of great value , and , in the meantime , we are justified in assuming that the temperature range values , as determined by us , are in such close agreement with the Paris thermometer as to render any alteration in our results unnecessary .
The reduction of our temperature from the air , to the absolute , scale would diminish the value of our total temperature range by nearly002 , but I do not consider the accuracy of our results of such an order as to render the resulting correction ( about 1 in 5000 ) more than a matter of form .
I have forwarded this note to Dr. GUILLAUME with the request that he would be so good as to add his opinion as to the probable limit of error of the observations with thermometer , together with any other comments which he think advisable .
His remarks are as follows:\mdash ; " " Le resultat fort remarquab]e des comparaisons faites par M. GRIFFITHS est sans doute le meilleur criterium de la limit d'erreur , tant des observations du thermometre que de son e'tude ; et l'estimation aurais faite a aurait ete sensiblement plus elevee .
En effet , dans ] resultst final viennent s'ajouter toutes les causes d'erreurs suivantes:\mdash ; * The range value on the corrected scale is ( in the case of ) , and this old diminish our value of by1 MR. E. H. GRIFFITHS ON THE VALUE OF .
Etude thermometre P. .
Calibrage comprise l'interpolation dans la supposition que la division est distant ) ; .
Determination du coefficient de pression ; .
Determination de l'intervalle fondamental .
Les corrections resultant de ces deux dernieres etudes n'agissent que sir le range : et sont de peu d'importance ( moins de degre ) sir un intervalle de 10 degrk Le , au contraire , peut introduire de point en point des erreurs J'ai montre autrefois*que , dans un calibrage fait avec tous les soins possibles par un observateur tres-exerce , l'extreme limit d'erreur , pour un thermometre divise en dixiemes de degre ' est de 001 ; le calibrage du thermometre a ete fait avec soin , mais par un methode abregee et l'on peut s'attendre a des erreurs de 002 environ sir les points determines directement ; le tube etant assez irregulier les des interpoles peuvent atteindre003 . .
Reduction thermometre azote .
Cette reduction se fait au moyen des tables generals deterl1linees par la faite un fois pour toutes , d'un group de thermometres en verre dur avec le metre a .
Dans l'application de cette table au thermometre , on suppose que corrections sont ] , c'est-A-dire que le verre et le mercure de ce sont identiques a ceux des thermometres-etalons .
Cette identite thermique a verifiee approximativement par des comparaisons precises faites sir un grand de thermometres ; mais si l'on songe : que le verre est un corps chimiquement complique , et dont dilatation est soumise a des lois singulieres ; ( b ) qu'il suffit difference relative de ( en valeur absolve ) dans le coefficient de dans formule de la dilatation lineaire verre pour provoquer un diflerence de 001 dans les indications du thermometre a , on no peut etre que fort etonne d'une tell concordance , qui resout la question vitale du thermometre a mercure , et detruit les prejuges que l'on avait contre cet instrument , aussi bien que sir la dilatation verre .
raison .
Celles-ci comprennent:\mdash ; les observations a diverses temperatures ; les determinations du zero .
On peut vraisemblablement s'attendre a ce que les erreu1s de rature , defauts de l'eclairage , les ularit menisque provenant en des * ' Etudes thermometriqucs avaux et Memoires du Bureau Iuternational des Poids et tome 5 , p. 2.3 ( 1886 ) .
THE MECHANICAL EQUIVALENT OF HEAT .
43.3 larites du tube , enfin les erreurs de l'estimation produisent au total des discordances degre au minimum dans ( a ) et dans ( b ) .
Toutes les causes mentionnees ci-dessus et no concernant que le thermometre auraient pu , sans aucune doute , conduire dans la reduction au thermometre a azote a des erreurs irregulieres atteignant 005 .
La difference de en venne s'explique suffisamment par deux causes:\mdash ; Pression capillaire moindre dans la colonne toujours descendante a zero que dams la colonne a menisque le plus souvent ascendant ou en equilibre moyen aux autres temperatures .
Dams les tubes de thermometres l'angle de raccordement vary entre et environ , ce qui , dans les tubes de 1 millim. de diametre , correspond a un variation de pression de 50 ] .
de mercure au maximum .
Le zero se releve des l'instant ou le thermometre est plonge dans la glace ; ce relevement au moment initial est a peu pres inde'pendant de la depression , et depasse degre par minute .
deux causes reunies suffiraient a expliquer des differences constantes plus fortes meme que celle qu'a trouvee M. GRIFFITHS ; pour la plupart des recherches , comme dans le cas actuel , else sont sans nulle consequence .
Il me sera permis en terminant d'insister sir la security que donne aux mesures de temperatures le control si indirect et en temps si parfait qui resulte des aisons de M. GRIFFITHS , et d'exprimer le voeu que la verification soit bientot etendue a un intervalle de temperature aussi large que possible .
Ju qu'a ce jour , l'etude du thermometre a mercure au Bureau International a ete faite entre et ; l'echelle sera prochainement etendue jusqu'a .
Le thermometre a toluene a ete parfaitement etudie et compare au thermometre a hydrogene jusqu'a ; a cette temperature la discordance entre divers instruments etudies individuellement n'atteint pas05 degre SECTION HEAT DEVELOPED BY THE STIRRING .
During our earlier experiments ( 1887 ) we stirred by eans of a screw which revolved about 200 times per minute .
Throughout 1888 and 1889 we continued to use the same form of stirrer , but endeavoured to make the mixing more efficient by increasing the rate of revolution .
As our experience increased we became strongly impressed with the necessity of thorough mixing .
Any slackness in this respect makes itself evident in changes in resistance of the wire , owing to changes in its rate of coolin irregularities in the radiation and conduction , probably owing to varying temperatures of the lid ; ( c ) irregularities in the thermometer readings ; ( d ) in the water equivalent , which then is found to alter to some extent with the mass of the contained water .
NDCCCXCIII .
4.34 .
E. H. GRIFFITHS ON THE VALUE OF One assumption underlies all the calculations , viz. , that whatever the mass of contained water , the water equivalent of bhe calorimeter remains constant .
* find this assumption is not correct , unless the water is thrown against every part of the calorimeter throughout each experiment .
At first sight it would appear that , whatever might be the " " lag\ldquo ; of any unwetted portion of the calorimeter when the floiv of heat had become steady , the quantity of heat absorbed by it for each of 1o C. would be the same , and that it would only be at the commencement of experiment that differences would be observable .
It must , however , be remembered that the radiation coefficient of the whole calorimeter would be altered by the ence of this colder portion ; and , as this cause of error would continue to hout the whole experiment , the effect would be appreciable .
In 1891 we adopted a new form of stirrer .
An ebonite shaft , passed down the central tube , and terminated at its lower extremity in an .
point , which on a metal bearing on the base of the calorimeter .
Two gold tubes joined together in the form of a V were fastened to this shaft , the apex of the V being downwards and immediately above the agate point .
This stirrer acted as a centrifugal pump , and , when revolving at a rate of about 1200 per minute , threw the water upwards and outwards with great rapidity\mdash ; the whole contents of the calorimeter being removed in a fevr seconds when the lid was off .
There could be no doubt that , provided was sufficient water to at all times cover the lower openings of the tubes , the was sufficient , and that no portion of the calorimeter could remain for any appreciable time at a different temperature to the water .
This form of stirrer appeared to us satisfactory that we did not attempt to improve upon it during our experiments 1891 ; and it was this misplaced confidence which was the chief cause of our in that year , to obtain satisfactory results .
Its defects may shortly be summed up as follows:\mdash ; ( 1 .
) As the depth of water in the calorimeter increased , the work expended forcing the tubes through the water also increased ; and the rate of rise in ture ( which ought to have diminished ) was much greater when the mass of the was large .
( 2 .
) This in the quantity of heat developed would not have been fatal , it could have been determined separately for each mass of water used , and we persevered in our attempts to thus determine it .
Unfortunately , however , it capriciously , even when the mass of water and rate of revolution were constant ; we have never arrived at any satisfactory explanation of these variations .
It probable , however , that their origin was due to differences in pressure between *If the water equivalent varies , the fact is rendered very evident when the values of deduced different masses of water are compared .
, in 1891 , we adopted forms of stirrers which threw against the roof of the calorimeter , our experiments were rendered useless by this constant souroe error .
The agreement between the values obtained iu 1892 is a satisfactory proof that this had been overcome .
THE MECHANICAL EiQUIVALENT OF HEAT .
agate point and its bearing , the slightest variation in the position of the shaft of the stirrer causing very different quantities of heat to be generated .
The quantity of heat generated by the stirrer varied from about one-sixth to one-fifteenth of the heat generated by the current , according to the electromotive force used ; and as the capricious changes above referred to amounted , in some cases , to 5 or 6 per cent. of the heat due to the stirring , they might affect our ultimate result by as much as a half and , in some cases , nearly 1 per cent. The quantity of heat appeared to vary with the rate of revolution in such a manner that was a constant , where was the time of rising 1o C. and the rate of revolution .
Our experience of this stirrer led to the following conclusions:\mdash ; ( 1 .
) The ideal stirrer must be able to throw the water from the bottom to the top of the calorimeter whatever the mass of the water .
( 2 .
) The " " work done\ldquo ; by the stirrer should be the same , whateyer the depth of the water .
( 3 .
) The " " work done\ldquo ; by the stirrer should be , as far as possible , employed in throwing the water upwards to the lid , rather than in simply causin a rotation of the water near the bottom of the calorimeter .
( 4 .
) The whole of the energy should , if possible , be employed in lifting the water and not in overcoming the friction between solid surfaces , i.e. , the bearings should be outside the calorimeter .
Although we found it impracticable to devise a form which would entirely fulfil the above requirements , the one adopted by us during our 1892 experiments was satisfactory .
During the spring of 1892 , we made a large number of trial experiments with fferent forms of stirrers , and it should be borne in mind the form it ultimately assumed was an example of the ' survival of the fittest A section will be found in Plate 2 , fig. 2 .
A cylindrical tube AB , 1 inch in diameter , closed at the lower end , rested on four little feet about 1/ 16 of an inch in length , which were fixed to the base of the calorimeter .
The plate at the lower end was perforated in the centre , so as to allow the end of the revolving shaft to pass through without touching it .
Round this centre hole were four slits through which the water in the calorimeter passed into the cylinder .
The end of the revolving shaft was fitted with a small cylinder of agate , which hung loosely within a ring fixed to the base of the calorimeter at .
The end of the agate was not in contact with the base of the calorimeter .
The bearing by which the stirrer was supported was fixed at the top of the central glass tube , and thus any increased downward thrust of the stirrer did not alter the friction within the calorimeter .
The stirrer consisted of an inverted cone with nearly vertical paddles at its extremities ( a section shown at ) .
When the stirrer revolved , the water was drawn in through the base of the cylinder , thrown outwards by the paddles , and , unable to escape at the lower end , mounted rapidly to , whence it was ejected iderable violence against the roof of the calorimeter .
If the rate exceeded MR. E. H. GRIFFITHS ON THE VALUE OF about -six revolutions per second , it was found that when the lid was the water was thrown many inches above the top of the cylinder , while the shaft remained dry .
A small window at caused a stream of water to be rown on to the top of the thermometer bulb .
The bearings at the top of this shaft ultimately assumed the form shown at small block of steel was fastened by a screw to the shaft , and the lower end of eel terminated in a cting ring , which revolved in a brass bearing of the shown at .
This brass bearing was , as already mentioned , fixed to the top of central glass tube .
A drop of oil placed in the channel , on the top of the brass , rendered this bearing practically air-tight , at all events , sufficiently so to prevent any possibility of evaporation .
* The shaft from to the top of the calorimeter was about 8 inches long , and made of ebonite ; the conduction of heat along this shah was , therefore , very small .
The heat generated in the bearing at passed almost entirely into the water of the outer tank , for the brass tube which terminated there was continued downwards to the steel lid , and thus the lower three inches were immersed in the water .
The tion at was at first shaped like a horizontal paddle-wheel .
It was , however , found that although it would work excellently for a time , a gradual accumu- lation of air at the centre of the paddle-wheel greatly dimimshed its efficiency .
Many attempts were made to surmount this difficulty , and it was only by the version of the upper portion into the form of an inverted cone that it was finally overcome .
To adjust the stirrer , the screw in was loosened and the shaft lowered until the end of the agate touched the base of the calorimeter .
The stirrer was then raised about 1 millim. , and fixed in that position by means of the screw at .
When investigating the action of these stirrers , the lid of the calorimeter was replaced by a sheet of glass which prevented the contents from being ejected .
As far as it was possible to see the interior through the shower of water that poured against the glass , the water without the cylinder assumed a slow circular movement only .
At the same time , there can be but little doubt that the entire oontents passed from the bottom to the top within the space of a second or two .
As long as the volume of the contained water exceeded 130 cub. centims. , the throw appeared to be the same , whatever the quantity , and , as the figures will show , the work done altered only slightly .
Our first series of experiments was performed with the object of ascertaining ( 1 ) under the same conditions , the work done by the stirrer was constant ; ( 2 ) In respect the work done varied , if at all , with the mass of water ; ( 3 ) If there was constant relation between the heat developed and the rate of revolution .
The results of our preliminary series are given in the following table , the rise temperature being expressed in millimeters of thermometer , since our object this time was not to determine the actual quantity of heat developed , but to gate the points mentioned in the above headings .
was found that this " " join\ldquo ; would supporb a of 7 or 8 inches of water without From the above results we drew the following conclusions:\mdash ; ( 1 .
) That under the same conditions the quantity of heat developed was the same .
( 3 .
) That the work done increased slightly with the mass of water .
( 3 .
) That was a constant quantity .
If Nos. 1 and 3 were established , No. 2 was of little consequence , as we proposed to deterlnine the heat developed for each mass of water .
When it is remembered that our " " rate ' , varied in the above experiments from 26 to 34 revolutions per second , the close approxinlation of to a constant is remarkable , and , as we proposed to maintain a rate of , as nearly as possible , 30 revolutions per second throughout our experiments , it was evident that any assumption based on this conclusion , could lead to no appreciable error .
Again , a considerable variation in the value of would have but little effect on the value of , the time of rising 1o C. These experiments were regarded as only preliminary , and no special care was taken with regard to external temperature or pressure , and , it is probable , that many of the apparent discrepancies arose from such causes .
Again , when the movement of tho mercury column was so slow that from 50 to 100 seconds were taken in millim. , it was difficult to estimate with great precision the time of apparent contact of the mercury column and the wire of the telescope .
Personal errors of this description must , however , in the run , tend to " " mean out\ldquo ; and give a true average .
Had we been to bring our mechanical arrangements to such perfection that the rate of revolution was absolutely constant , it would have rendered the above investigation unnecessary , for the correction would have been constant whenever the mass of water was the : am .
Although by the pressure regulator , described p. , the regularity of the motor was increased to an extent which those accustomed to the behaviour of these instruments hardly anticipated , we were unable to ignore the residual variations .
It was , therefore , necessary to record the revolutions ( the time per 1000 being taken in each case ) throughout all our experiments .
The labour , both of observing and calculating , was thereby greatly increased ; but as we had no better means at our disposal , we had to submit to the inevitable .
The establishment of the law that ( with this stirrer ) might be regarded as constant , enabled us to apply with certainty the necessary corrections .
As previously stated , our object was to deduce the rise in temperature per 1 second , rate 30 .
If is time of rise of in any scale , then , therefore , @ . .
Sr. Knowing ( the rise per 1 second at any observed rate ) , and ( the rise at rate 30 ) , we know , that is the term required to reduce the observation to the rate 30 .
The accuracy of the correction can be estimated by means of the above table .
For time rising ( in any , in seconds .
rate of revolution , i.e. , numb of revolutions per second .
THE MECHANICAL EQUIVALENT OF HEAT .
example , the mean value of for the weight is 1600 , the greatest departure lrom the mean is 1611 in Experiment 13 .
By assuming , we find the rise per 1 second at rate ( expressed in millims. of ) , whereas if we assume that and correct Experiment rate to rate 30 , we get I676 .
* Again , the mean value of for weight 28537 is 2119 , which gives the rise at rate 30 as001274 .
Experiment 22 rate is an extreme case ; this would give at rate 30 .
Experiment 7 ( weight ) shows a greater divergence from the mean value , since it would give a rise per 1 second as against 00238 , the value as given by the mean .
As , however , we discarded all experiments where the weight of water was less than 130 grms. , this extreme case did not affect our results .
During our experiments our rate was always between 29 and 31 ( generally between and ) , hence the divergences in the cases , above selected , are far greater than any which occurred during our experiments .
The value of appeared to increase very regularly as the quantity of water increased , and thus charlges in its value caused by small changes in the mass could be deduced .
Thus experiment 30 grms. gave a value of 18,39 .
Hence we deduced that , if the weight of water was 23973 ( as in Nos. 17 to 19 ) , the value of as against 1842 , the mean of the actual nents with the latter weight .
Had the work done by the stirrer been independent of the quantity of water , we could have deduced the water equivalent of the calorimeter from the values of , for , if , etc. , are the values of when are the weights , we have as the value of the water equivalent .
Usin the values of deduced from Table XVIII .
, we get We ultimately ascertained that the water equivalent was as nearly as possible .
Assuming this value and denoting by the mass of contained water water equivalent , then , if the work done by the stirrer was constant , would be constant .
The numbers in the last column show that the quantity of heat develcped increases rapidly at first and arrives at a maximum .
maximum , as well as we can estimate , occurs when the quantity of water is * We required but three significant figures when we applied a similan ction to our experiments .
sufficient to fill the calorimeter to about the top of the tube uP which the water thrown , and it is possible that the conditions would , at this depth , change slightly .
In any case , the work done by the stirrer alters but very little as the height water changes , provided that 130 cub. centims. is regarded as the minimum quantity .
The examples given above show that the limit of error introduced by applying method of reduction to variations far exceedin any occurring during our experimen is less than 1 per cent. , and as the heat generated by the stirrer during the majority of our 1892 experiments varied from about to of the heat due to the current , is evident that any error must be less than of the whole supply .
Bearing mind the small variations which actually occurred , we may safely regard the limits error as a small fraction of this As will be seen later on , we performed an experiment in which the current so diminished that the stirring supply was half that due to the current ( see ) , and another in which the current was so increased that the proportions were as 1 : 20 see .
Any inaccuracy in the determination of the stirrin supply would have caused considerable divergence in the corrected results ; but the numbers obtained are themselves a proof of the accuracy of the method adopted .
See note , p. 406 .
) It appeared possible that changes in the viscosity of water at different temperatures would have an influence on the results .
This may be the case , although our experiments do not lead to that conclusion ( see p. 448 ) .
If so , however , it would not affect us , for the total heat lost or gained by the calorimeter at difl'erent temperatures ( independently of the current supply ) was directly determined for each weight of water , and therefore although our values of the radiation \amp ; c. , might be modified , the expression for the total loss or gain would remain unchanged .
Want of time alone prevented our ating this matter more fully , but the consideration that neither the value of , nor the changes in the specific heat of water would be affected thereby , justified its postponement to another season .
No use was made of the numbers obtained in Table XVIII .
; our only object being to establish the law .
This was fortunate , for on afterwards examining the water used for this series we had reason to believe that it had become slightly soiled .
We found that the lead ring used in the calorimeter lid had been insufficiently gilded , and also we had not allowed the gutta-percha varnish , which we used as a " " size for the gold leaf , to harden sufficiently before introducing the water .
The results of nearly ninety additional experiments of a similar kind indicate that the irregularities of the numbers in the above table are unusual , and that the actual values of , obtained from them , are unreliable , but that the conclusion drawn that under same circumstances constant ) is correct .
Very small changes in the adjustment caused considerable changes in , but *This minimum quantity , 130 centims. , is deduced the results of later periments .
THE MFCHANICAL EQUIVALENT OF HEAT .
value remained constant so long as the adjustment was unaltered .
Any alteration in the fixing of the stirring shaft necessitated a re-determination of K. After experiment , the whole apparatus was taken to pieces .
It was then found that small fragments of gutta-percha had , by the continued impact of the water , been detached from the ring of gutta-percha which had been placed between the cooling tube and the tube down which it passed .
Many of these had collected beneath the base of the cylinder at and must have considerably impeded the flow of water .
In consequence , therefore , before putting the apparatus together again , we removed all remaining gutta-percha surfaces which were exposed to the action of the water .
The values obtained for were , from this time , very rent from preceding ones , and , once obtained , remained constant as long as the same mass of water was used .
The experiments to 34 , are in better agreement amongst themselves than the earlier ones , and the cause no doubt is thatjust indicated .
The mean result of the earlier experiments is satisfactory as likewise the mean time of each experiment , but the times over small ranges ( especially in experiments Nos. to 12 leave much to be desired .
The fact that the quantity of heat developed by the stirrer was increased by the removal of the gutta-percha fragments is a proof that the flow of water was appreciably diminished by the obstruction .
SECTION \mdash ; THE GAIN 0R Loss BY RADIATION , \amp ; c. It has already been pointed out that the method of observation adopted enabled us to determine , with sufficient accuracy , the total loss or gain of heat per second at any temperature , apart from the supply due to current .
We propose to use the phrase " " non-electrical supply\ldquo ; to denote heat due to all other sources than the current .
By the method described in Section XI .
, we were enabled to deduce the rise per second when the rate of revolution of the stirrer was , although the actual rate differed slightly from our standard one .
The rate of rise , when radiation , \amp ; c. , was eliminated , was first xined by time of a small rise across the outside temperature .
The temperature of the calorimeter was then gradually raised through the whole range by means of stirring only , and the times ascertained of small changes at the different temperatures .
Assuming the ' stirring heat\ldquo ; to be the same at all temperatures ( when the rate is the same ) , we can deduce the radiation ) efficient .
( By radiation coefficient we denote the loss or gain in temperature per second , due to the combined effects of radiation , conduction , and convection , for a di{ference in temperature of 1o C ) .
One advantage of this method is , that all observations of telnperiture taken , fig. 2 .
[ .
\mdash ; A. a rising thermometer , for " " sticking\ldquo ; is far less likely to occur when the mercury advancing than when receding .
We have found that all observations of slowly temperatures ( when taken with a mercui.y thermometer ) are unsatisfactory for the above reason .
Again , conditions resemble more nearly , in every respect , those prevalent during a \ldquo ; experiment .
On this account , we did not desire to reduce the " " stirring heat even We had been able to do so , since the loss , due to radiation , \amp ; c. , would , at the portions of the range , have mastered the stirring supply , and the observations have had to be taken on a falling thermometer , a mode of observation which , by the results obtained in previous years , we had decided to reject .
It appeared unlikely that NEWTON'S Law of Radiation would hold our and we anticipated a decided curvature in the line showing our ] supply We first proceeded to determine isolated points on this curve ( ordinate rise in temperature per second , abscissa difference in temperature between the calorimeter and the outside temperature ) , but these appeared to fall so nearly on a straight line , that we decided to follow its course over our range of temperature , using a small mass of water so that the changes in the value of the rise should be as marked as possible .
A very large number of experiments were made in order to decide this point\mdash ; for , if the exponential formula given by DULONG and PETIT holds , the curvature ought to have been marked .
Although changes in the specific heat of water and of the calorimeter would influence the inclination of the line , it appeared unlikely that they would affect it in such a manner as to straighten it ; and further , the effect of such changes would alter as the mass of water changed .
* The numbers alone resulting from these observations half fill a large note-book , and to give them in full would occupy too much space .
As an example , we venture to give experiments 51 to 54 , exactly as they appeared in our notes .
As the method employed never vairied , we will , in other cases , simply give results , and we select this series because it was the first done by us with the object of tracing the ' non-electrical supply\ldquo ; curve throughout the greater part of our range .
The numbers in the columns marked ' time are those given by the chronograph tape .
New tapes were used after the readings at 343 , 407 , and , and the times of rising between these temperatures were not noted ; therefore , the experiments at those ranges are numbered as separate although taken consecutively .
No temperatures were taken whose readings ended in or 5 because , as these were *We may state at once that ( altbough some slight signs of ur are visible when the mass of water is greab ) assumino .
tho non-electrical supply to be a linear nction of the temperature ( when the rate of ) , we introduce no error sufficient to affect our results , and this clusion is based on the data supplied by more than 100 ' stirring and radiation \ldquo ; experiments .
If the Ubbed in of the ) mometer s , tho curvature would , of course , be THE MECHANICAL EQUIVALENT OF HEAT .
TABLE XX.\mdash ; Experiment LII.-LIV .
Aug. 11 , 1892 .
The constants are the same as for LI .
Method of reducing the results in the The time over any range is the mean time obtained by taking the observations in pairs\mdash ; thus the time over the range 875-97 is the mean of the times taken in rising from 86-96 , 87-97 , 88-98 , and 89-99 .
Thus each number in column 4 of the Table is the mean of , at least , four intervals of time .
THE MECHANICAL EQUIVALENT OF HEAT .
The outside temperature throughout the above experiments was C. The actual observation across that temperature was over a range of C. only , and an error of C. in the range would make a difference of 4 in the number in the last 6olumn .
As the rise at this point was one of the most important of all , a straight line was drawn through the numbers obtained from the pairs of observations on each side of it , the resulting value bein .
Assuming this to be the rise due to stirring only ( at rate 30 ) , we get radiation coefficient , where is rise ( gain ) per second due to all non-electrical sources .
The mean temperature of the first four observations is C. , and the mean value of , hence The straightness of this " " non-electrical supply line\ldquo ; is a matter of extreme importance , as if there were any marked curvature it would seriously affect our con- clusions as to the specific heat of water .
In the last columns of the above Table we give the numbers obtained by direct calculation , assuming and , also the difference between the calculated and experimental results .
The irregularities are obviously experimental ones , and are to be expected when it is remembered that an error of C. in the " " range\ldquo ; would account for nearly all of them .
Again , we are here dealing with our smallest mass of water , and the individual experiments are in better agreement when the depth of water is greater .
The even distribution of the numbers about a straight line is rendered more evident to the eye when plotted on a large scale than when presented in rows of figures .
The Table also illustrates the accuracy of the correction for differences in rate of stirring .
* Assuming the value of the water equivalent of the calorimeter as , the value of ( mass of water water equivalent ) , we get as the number of thermal grarnmes lost or gained by radiation , \amp ; c. , per second for a difference in temperature of 1o C. In order to test to what extent any change in the viscosity of the water caused a change in the stirring heat , we performed a few experiments with different external temperatures .
* This electrical supply CUl.ve\ldquo ; being a straight line , wo are enabled to deduce the valuos of both and by experimenbs only , conducted at ally two different mpcratures .
MR. ] ] ] .
H. GRIFFITHS ON THE VALUE OF TABLE XXII .
As we regarded these experiments only as preliminary , they were somewhat hurriedly performed , and we proposed investigating this point more fully later on .
It has been previously pointed out ( p. 440 ) that , in any case , the conclusions arrived at would not affect our results , and we only give them to show that no marked change is indicated .
Some similar experiments performed in 1891 had led to the same conclusion .
During this year 's ( ls92 ) experiments we were not troubled by any air , or water , leakages into the space between the calorimeter and the steel chamber .
great portion of our time and attention in previous years having been devoted to leak-hunting , we had considerable ( unwished for ) opportunities of investigating the effects of changes of pressure , or alterations in the dryness of the surrounding medium .
The smallest trace of moisture hss a most ishing effect on the radiation coefficient .
The change in the rate of rise is most marked , and if the moisture present is sufficient to saturate the space when the calorimeter is cooled down , all attempts to obtain a radiation coefficient are hopeless .
Any trace of aqueous vapour could immediately be detected by its effect on the rate of rise , and on one or two occasions was sufficient to prevent any rise whatever .
Considerable effect was also caused by changes in pressure , and when comparing the results tained from different masses it must be remembered that the pressure was frequently changed .
Although we were then not conscious of the full ance of this point , we endeavoured to maintain the pressure unaltered from the time that any of stirring determinations were made until the corresponding Jexperi llents were coml ) leted .
CHANICAL EQUTVALENT OF HEAT .
Results of the and Radiation Experinents .
Throughout the remainder of this section the following notation is adopted:\mdash ; weight of water ( in vacuo ) .
water equivalent of calorimeter .
pressure in annular space .
time of rising 1o C. rate of evolution of stirrer ( revolutions per 1 second ) .
value of rise per 1 second ( in degrees C. ) due to stirring only .
gain or loss ( in degrees C. ) per 1 second due to radiation , conduction , and convection , when the difference of temperature C. thermal grms. lost or gained per 1 second by radiation , conduction , and convection Experiments 1 to 31 ( see Table XVIII \mdash ; These were all observations taken acros the outside temperature only , and were used to establish the relation K. Experiments.32 to 40.\mdash ; Are only of use for the same purpose , but they bear out the conclusion that K. As the calorimeter was taken to pieces after Experiment 40 ( before any of our experiments were commenced ) the results are of no special value .
Experiments 41 to 50:\mdash ; millims. Experiments 51 to 57.\mdash ; ( Tables XIX .
, XX .
, and XXII .
) ( Nos. 41 to 54 were used in the reduction of to 4 .
) Experiments to 65.\mdash ; During this series the value of began to show signs of alteration .
The cause has already been mentioned ; flakes of gutta-percha having collected at the base of the inner cylinder , the ' throw\ldquo ; of the stirrer became less regular .
Also ( as we were not at this time aware of the importance of maintaining the pressure unaltered ) , we both diminished and increased the pressure during this series .
Fortunately , however , we performed sufficient radiation expel.iments to enable us to deduce the probable values of the variables appropriate to the respective determinations .
throughout this series .
Experiments 58 and 59:\mdash ; millim. , Experiments 60 and 61 :\mdash ; and 9 .
millim. Experiments 62 to 65:\mdash ; to 12 .
millim. Experiments 66 .\mdash ; The cause of irregularity had not yet been ascertained , and therefore not removed .
Durin this series we made several changes in the connections between the stirring rod and the revolving shaft above it , and also in the pressure around the calorimeter .
throughout this series .
Experiments 66 to 71:\mdash ; and 14 .
millim. Experiments 72 and 73.\mdash ; The position of the revolving shaft and counter was slightly altered , and the whole of this portion of the apparatus re adjusted .
and 16 .
millim. Experiments 74 to 78.\mdash ; The double HOOKE'S joint was now removed and replaced by a single one in the hope of diminishing the vibration of the stirring rod , the pressure was at the same time somewhat reduced .
The value of showed considerable alteration .
and 18 .
millim. * Therc is some here .
We have only a note of the pressure during this ( August 28 ) , and we have no means of checking it since Experiment 74 was not performed unffi August 30 .
Now ( as we find recorded in our notes ) , we gave twenty-three strokes to the pumps on August 29 , the resulting pressure on August .30 being 37 millim. ; it appears certain , therefore , that the pressul.e on August 28 must have exceeded38 millim. THE MECHANTCAL EQUIVALENT OF HEAT .
Experiments 79 and 80.\mdash ; The simple HOOKE'S join was now replaced by a light l.ing of wire , which passed through holes in the stirrin rod and the revolving shaft .
millim. It thus appeared that both and had resumed their formel values .
Experiments 81 and 82.\mdash ; These two experiments were intended to be the first pair of a series .
On commencing , the same evening ( September 4 ) , it was discovered that the coil in the calorimeter had become loose .
The calorimeter had , in consequence , to be taken to pieces , and a week elapsed before we were able to resume our determination .
'rhese two experiments are consequently of no use , but they are interesting as showing the kin of accuracy obtainable .
millim. Experiments 83 to 93.\mdash ; After Experiment 82 a marked diffel.ence is observable in the value of .
Between September 4 and 10 the calorimeter had been thoroughly deaned , and all gutta-percha surfaces removed from its interior .
The " " throw\ldquo ; of the stirrer was evidently improved , and as a consequence the value of was raised .
The series of J Experiments Series , performed after this change , are so much superior to their predecessors that we base our conclusions almost entirely upon them .
We feel it right , therefore , to give more particulars concerning their " " non-electrical supply\ldquo ; determinations than we have considered necessary in the case of Series I. ( to 19 ) .
The following table is a summary of the " " stirrin experiments\ldquo ; used in the reduction of to 25 .
TABLE XXIII .
THE MECHANICAL EQUIVALENT OF HEAT .
Hence millims. The reduction of our observations involved so much arithmetic that the whole series of experiments was concluded before we were able to get results in the form now given .
Although we were prepared to find a rapid reduction in the value of as the pressure diminished , we had not realized the importance of small changes in the pressure .
This was unfortunate , for , as the figures show , our average pressure was just that at which small alterations produced the reatest effect .
We usually considered it sufficient to record the pressure each day , whereas it appears that a careful observation of the pressure during each stirring experiment ought to have been made .
It is evident that we should have done better to select pressures of not less than 1 millim. , and thus small variations would have had but little effect .
At the time of the observations we did not know the real value of the water equivalent .
It was thus impossible for us to obtain with any accuracy , and it is only by the changes in the value of that the effects of changes in pressure are rendered evident .
The results compare very favourably with the conclusions arrived at by BOTTOMLEY .
* Collecting the numbers given in this section ( with the exception of Experiments 72 and73 , see note , p. 450 ) , we get TABLE XXVI .
These numbers are in fig. 8 the )scissae denoting pressures and the ordinates the number of the thermitl grms. gained per second .
When it is remembered that adjacent points are obtained from very different masses of water , the close agreement a of the accuracy of the observations .
The sudden change in the value of ' Phil. Trans 1887 , A. when the pressure falls below .
millim. is remarkable ; but a reference to BOTTO paper will show that the curve given by him is of a similar type , although critical point is at a higher pressure and the bend in the curve is less sharp .
Fig. 8 .
Although all the observations on the " " non-electrical supply\ldquo ; have been brought together in this section for the purposes of comparison , they were not taken con : tinuously but in groups .
Each group of observations was used for the reduction ot experiments performed at the same time and on the same weight .
The stirring experiments were usually performed during the daytime , the corresponding experi ments in the evening , and it was rarely that two sets of observations were taken on the same day .
Throughout the whole of the experiments to 34 , as also their corresponding stirring experiments , the mding temperature was maintained at If temperature of calorimeter , then gives the rise or fall in temperature per second due to radiation , \amp ; c. , hence , gives the rise in temperature per second due to the whole ' non-electrical\ldquo ; supply .
previously pointed out ( see note , p. 411 ) we oroposed to reduce the numbers obtained from our experiments at certain fixed temperatures and the following Table giv the changes in temperature per 1 second , due to the ' non-electrical " " supply at points .
The numbers were obtained from large-scale curves , but can be calculated from the values of and given in this section .
* In previous years we had performed several of experiments in which the outside temperatu was , in one case at the bottom , and in the other at the top , of our range .
By " " meaning\ldquo ; the reciprooa of the times of rising , the effect of , \amp ; c. , could be eliminated .
We hoped to repeat experiments this year , but want of time ) revented our doing so .
The above Table is deduced from " " \ldquo ; experiments Nos. 41 to 110 .
Th values of and , obtained by repeating an experiment under the same rarely show any difference as great as 000004 , and thus a difference from the mean value of 000002 may be regarded as our limit of experimental error .
Now a changS of two in the last digit would , when 5 Clark cells were used , cause a difference than 1 in 4000 in the resulting value of ; but as the numbers in each column are obtained from independent experiments , and as all the columns are used in the final determination of , it is probable that the mean resulting error is of very small dimensions .
[ Note by E. H. G. Added April , 1893 .
statement that the loss or gain by radiation , convection , and conduction is , in our apparatus , a linear function of the difference in temperature may at first sight appear to differ from the conclusions arrived at by DULONG and PETIT , MACFARLANE , and other observers , but I think that the contradiction is more apparent than real .
The curve which shows the absolute loss or gain ( fig. 8 , p. 454 ) indicates that , if we had reduced the pressure to something under one-thousandth of a millimetre , the loss or gain by radiation ; convection , and conduction might almost have been disregarded over our range , and hence it is evident that convection plays by far the greatest part at such pressures as those with which we were working ; and the wonderful experiments of Professor DEWAR ( the account of which had not been published when the preceding section was written ) appear to place this conclusion on a firm foundation .
Now DULONG and PETIT make the loss or gain by convection vary directly as the difference in tempera .
ture when the density of the gas remains constant .
In our apparatus the density of the air surrounding the calorimeter must have remained nearly constant throughout each experiment , and therefore our observations appear to corroborate , rather than to contradict , the conclusions arrived at by DULONG and ] SECTION XIII.\mdash ; THE METHOD 0F PERFORMING A EXPERIMENT .
The general method of procedure was as follows .
Freshly distilled ether having been placed in the cooling tube , the temperature of the calorimeter was lowered until the resistance of the coil indicated a temperature of C. , any remaining ether was then withdrawn and dried air passed through the cooling apparatus for about ten minutes .
The aspirating portion of the apparatus was removed and the cooling tube cleaned by a small mop of absorbent cotton wool .
The open end was closed by a long cork , and a metal tube placed over the whole of the projectin portion , the lower end of this tube being washed by the tank water .
The storage circuit was then completed and the rheostat adjusted until the balance with the Clark cells was obtained .
About twenty minutes elapsed before the temperature rose to 86 \mdash ; the point at which we commenced our observationS .
During this interval the chronograph circuit was tested , and any final adjustment of the motor and external circuit completed .
THE MECHANICAL The task of one observer was invariably that of maintaining the potential balance by means of the rheostat , and also inverting the currents at regular intervals .
The attention of the second observer was directed to recordin the gronp of transits about each fixed point , the chronograph key bein placed at the base of the telescope stand .
In the intervals he recorded the times of the revolutions of the stirrer , which were distinguished on the tapes from the temperature records by double marks .
He also occasionally recorded the readin of thermometer ( the external temperature ) , but these were as a rule so regular as to render this observation a matter of form .
Thus , during the progress of an experiment , no notes had to be taken unless some exceptional incident occurred .
We believe this to be an important matter during observations of this description , for the mechanical operation of note-taking greatly distracts the attention .
The chronograph required rewindin about every twenty minutes , and in the press of the other observations this was sometimes forgotten .
Some of the omissions in he tables are due to this cause .
The duration of a whole experiment varied from 40 to 80 minutes , according to the weight of water present and the E.M.F. used .
At the end of an experiment , the chronograph tape was counted and the results tabulated in the form shown in Tables XXVIII .
to XXXIII .
These six experiments are typical of the rest , and have not been in anyway selected , except that they are the first performed on each weight , in Series II .
, with different E.M.F. THE MECHANICAL EQUIVALENT OF Observations on XXX V. XXX VI . .
to 4 .
These have been rejected and the observations are not included in the Tables .
The discrepancles over the shorter ranges are absurdly great , and the mean times far too large .
During these experiments we had great difficulty in maintaining the potential-balance ; this ought to have been sufficient to indicate the source of error , but , at the time , we attributed it to want of practice in the observers .
It must be borne in mind that it was not until many weeks after the close of our observations that ions were completed , so that , at the of experimenting , we were wholly in the dark as to the actual value of our results .
Our only means of obtaining any glimpse of the value of a particular experiment was its resemblance to some othel one taken under similar conditions , but , as all the conditions were never precisely the same , we could bhus obtain little information .
We have now no hesitation in assigning the following cause to the failure of these four .
With this weight of water , the depth in the calorimeter , when the stirrer was at rest , was only 2 centims. , and as , when rapid stirring took great portion of the contents must have been within the cylinder and in the upper portion of the calorimeter , it is evident that the higher parts of the coil must have been at times uncovered , since it required centims. for complete immersion .
This would account ( 1 ) for the irregular behaviour of the potential-balance , and ( 2 ) for the great lengthenin of the times shown in these experiments .
We at no time antici- pated success with so small a mass of water , but the excellent results obtained from our stirring iments on this weight ( see Table XXI .
) justified the attempt .
Experiments 5 to 12 .
( Group ) These experiments only slightly more satisfactory than those we have rejected .
The cause is partly that so frequently alluded to in Section XII .
, viz. , obstructions in the openings at the base of the internal cylinder , the resulting irregularities in the throw of the stirrer affecting not only the non-electrical supply , but also changing the coefficient of increase in resistance of the wire .
That this is the case is indicated by the results of Experiment 9 , where the E.M.F. was so reduced that the current was only .
ampere , and the changes in the temperature of the wire were less marked .
This experiment was performed in order to test the accuracy of our various corrections , rather than as a serious effort to determine the value of J. The oservations in this only extended over of our total range , for , had we endeavoured to it out through the upper ranges , its duration would have extended over some hours .
In every case ( excepting as above stated , Experiments 1 to 4 ) we given the numbers recorded , because , however bad individual observations may appear , we have no sufficient grounds for their rejection .
The omissions in the table represent experimental mistakes ; in No. 5 , Range 7 , the storage circuit We'lS broken a second ox two 1oy an accident with the reversing key ; in No. 6 , the chronograph refused to work hout the first two ranges ; in No. 10 , some contact must have taken plaoe between the wires of the coil , as the potential-balance was entirely destroyed the 7th .
No. 11 was exceptionally bad experiment , especially in the latto$ half ; the changes in water pressure were so rapid and violent as to pass beyond ths control of the motor-regulator , the rate changing from 2943 to in two consecu .
tive observations .
Had it not been for our desire to present our results in full , would have rejected this experiment .
to 19 .
In this group , our heaviest weight , the observations although still somewhat irregular are in better reement .
As might be anticipated the throw of the stirrer was less affected by the obstructions at the base when the depth of water was greater .
The alterations made in the connections of the stirring rod have been already alluded to .
Our conclusions as to the value of and the changes in the specific heat of water are luased almost entirely on the results of Series II .
Although the results deduced from Nos. 13 to 19 are in almost perfect harmony with those from Series II .
, the agreement , for the reasons given above , must be in part fortuitous .
The mean value of deduced from Nos. 5 to 12 agrees with the most probable value , but the change in the specific heat of water alters from positive to negative according as the group is compared with the results obtained from a lighter or a heavier weight .
Series II .
This series contains but fifteen experiments , which may appear a small number on which to base our conclusions .
We would , however , point out that these are only the concluding series of many hundred experiments conducted by the same observers , who are therefore in the best position to judge of their value .
more , these fifteen experiments are conducted on very different weights of water and with very different currents , and our conclusions would be but little altered if they were based on any two individual experiments rather than on the mean results of alh Want of time alone prevented our increasing their number , but we believe that such increase would have given them small additional weight .
The results of Series I. are quite sufficient to corroborate the value of the least trustworthy group of that series gives contradictory values for the change in specific heat of water .
Such irregularities as are observable in the members of these series , are we inseparable from the use of mercury thermometers .
Although many irregularitiel which would otherwise be apparent have been eliminated by the method of calibration adopted ( Section X there yet remain residual ones , not strictly recurrenk whose effect can only be eliminated by taking the mean path of a series of obser vations over consecutive ranges .
The magnitude of such irregularities can THE MECHAINICAL EQUIVALENT OF HEAT .
stimated by a comparison of the columns headed " " mean\ldquo ; and " " from curve\ldquo ; in fable None of these experiments call for special comment , with the exception of No. 34 .
was performed with the object of testing the accuracy of the correction given by formula SR ( Section VII .
, Table IX and we did not anticipate from it results of much value .
The irregularities in ranges 8 , 9 , 10 , of Group , Column 21 , able X , are clearly an example of the eccentric behaviour of the thermometer when rising this rapid rate ; their mean however is excellent .
SECTION XIV.\mdash ; THE ALCULATION 0F THE RESULTS The method which we have adopted in our calculations can be put into a general form , thus : Suppose to be the rate of production of heat in the calorimeter at some standard temperature , whilst denotes the value of at any other temperature ; the change in temperature per one second due to radiation , \amp ; c. , when the difference between the external and internal temperature is 1o ; the capacity for heat of the calorimeter and its contents at the standard temperature its value at ; Then if is the temperature of the calorimeter at any time , the rate of production of heat is , where is the temperature of the surrounding envelope .
Thus the equation of condition is Now , by JOULE'S law .
JM ( 2 ) , therefore ( 3 ) , where is the rate of production of heat due to the eJectrical suppJy , and is the E.M. .
of a Clark cell , and is the number of cells used .
But since the rate of production of heat is dependent on the resistance of a platinum wire , and is kept constant , this rate will diminish as the temperature rises , or , . . . . . .
( 4 ) , We have changed the notation adopted in Section I. to one more suited for arithmetical operations .
where is the resistance coefficient of the wire , and is the number ot degrees that the wire is hotter than the water in contact with it .
Similarly where is the meall coefficient of increase of specific heat of the water and calorimeter .
For the 11011-electrical supply , we have shown that ( p. 438 ) , where is the time of rising 1o C. and is the rate of revolution of the stirrer , and therefore th6 rise per second is given by . . . . .
( 5 ) , and S . . . . .
( 6 ) .
By ( 6 ) we can reduce all observations to the standard rate Equation ( 1 ) after introducing terms due to stirring becomes thal is , ; or .
If we write and for the electrical and non-electrical terms in this expression , ( 9 ) .
tion ( may be expressed in the following form\mdash ; where A is the sum of quantities !
.ndependent of , and is the sum of a number of small quantities , The range is ) ckoned from slightly below , say from , to slightly above , say ; hence if and are the times at ' and we may assign to the value Strictly speaking , the equations ought to be integrated from , when , to when , but the error introduced by the above method may , provided the ranges are sufficiently mall , be neglected .
We have adopted throughout experimenffi ranges of from to C. It does not appear prob able that increassd accuracy would have resulted from the adoption of smaller ranges , for the effect errors in the thermometry is increased as the size of the ranges is diminished .
THE MECHANICAL EQUIVALENT OF HEAT .
The nents on the heat developed by the stirrer give us values of ( See Table XXVII .
) 1 .
The first step in the calculation of our observations has been to correct , where necessary , all times .
the clock error , which was , up to inclusive .
This correction has been applied in Table XXXVIII .
2 .
We found the value of , i. e. , rise per second produced by the combined effect of all sources of gain or loss ( Table XXXVII .
, col .
4 ) .
3 .
Since the rate of revolution throughout an experiment was always nearly 30 per second , we eliminated irregularities in the rate by reducing rate 30 by means of the formula given in ( 6 ) .
Thus we obtained the value would have had , had the rate been uniformly 30 per econd .
This reduction is shown Table XXXVII .
, cols .
5 and 6 .
4 .
In order to be able to more readily compare experiments performed under almost identical circumstances , the next correction applied was that for the telnperature of the Clark cells .
Lord RAYLEIGH gives the formula hence . . . . .
( 11 ) , where is temperature of the Clark cells .
This correction oannot be to as a whole ( equation g ) but only to the portion the electrical supply .
As the correction was always very small , it was sufficient to find the approximate value of by subtracting , any temperature , the value of as given in Table XXVII .
, and then find the correction on the yalue of so obtained by formula ( 11 ) .
The result of this correction gives col .
7 , Table The correction was applied in this order as we could thus compare all the experiments performed with the same E.M.F. 's so long as the values of and were the same .
The results at this stage of their reduction are given in Tables XXXVIII .
and XXXIX .
5 .
XXXVII .
, col .
s , gives the value of taken from Table XXVII .
This tracted from col .
7 giyes the value of , i.e. , the rise per second due to the current only , the difference of potential at the ends of the coil Clark cells ( col .
9 ) .
Hence equation ( 8 ) is reduced to 6 .
In the section on resistance in Table XVI .
we have collected the values of at the mean temperature of each range .
The value of number of Clark used , is given with each experiment .
Substituting their we obtain the time of rising 1o C. with unit resistance and one Clark cell , or ( 12 ) .
This gives column 10 of Table XXXVII .
7 .
The results of all the experiments were then meaned and plotted , the columns " " Mean\ldquo ; and " " From curve\ldquo ; in Tables .
and XLI .
giving the results .
From these smoothed curves the values of the required constants were calculated thus : 8 .
If is the water-equivalent of the calorimeter , and and are the temperature coefficients of the specific heat vater and of the calorimeter respectively , and suffixes written to , denote the values of these quantities for different weights of water , then from ( 12 ) we obtain and ( 13 ) , where and ( 14 ) .
Hence , subtractin equations , we get or and ( 15 ) , THE MECHANICAL EQUIVALENT OF HEAT .
where the obseryations are taken at two temperatures and , the second suffix of the T'sreferring.to these temperatures .
Hence , by division , we obtain the value of , the temperature-coefficient of the ecific heat of water .
And , if is our standard temperature , the value of is found without any knowledge of the water-equivalent of the calorimeter .
Now is the time that grms. of water at temperature would take to rise C. cubtracting this from and dividing by we obtain the number of grms. of water at , to which the calorimeter also at is equivalent . . . . .
( 16 ) .
Similarly the expression ( 17 ) gives the number of grammes of water at to which the calorimeter at is equivalent .
Thus we obtain both the water-equiyalent of the calorimeter and also its temperature-coefficient Finally is calculated from .
NDCCCXCIII.\mdash ; A. THE MECHANICAL OF HEAT .
Remarks on The column headed " " Mean\ldquo ; was obtained by multiplying the numbers in the preceding columns of each section by the number of experiments from which they were deduced , and dividing the sum of the products by the total number of experiments .
The " " means\ldquo ; were plotted on a scale such that inch of ordinate repre- .
sented a difference of 1 second in , and the smooth curve in each case so drawn the sum of the positive and negative areas included between it and the experimental curve was zero .
The curvature is so slight over our range that by assuming t be a linear function of , our " " curve numbers\ldquo ; are not affected when written to four figures .
It will be found that our results ( obtained from the " " smooth curve\ldquo ; values of T ) are in much closer reement than might be expected from a study of Table This indicates that the irregularities are due to some ( or all ) of the following causes .
( 1 .
) Errors in the coml ) arative values of the ranges .
The effect of such errors would be common to all the experiments , and there is evidence that the sixth ange is too small , the value of over that range being nearly always too large ; the errol however , does not amount to .
Again , the ninth range is evidently too great .
These errors do not affect the result when the values of over two or three consecutive ranges are meaned ( as is done by the smooth curve ) , the excess in one case is then compensated by the deficiency in another .
( 2 .
) The ( as apart from the strictly recurrent ) " " stickings\ldquo ; of the thermometer ; these , as previously mentioned , must mean out when sufficient observations are taken over the same range .
( 8 .
) ularities in the temperature of the water ejected through the in cylinder on to the thermometer bulb , water from cooler parts of the calorimeter being followed by a gush of warmer water brought direct from the hot wire .
This , no doubt , is a fruitful cause of alternate lag and acceleration , but , it is evidently an irregularity which would but slightly affect the results of many continued through a sufficient number of intervals of time .
We have previously pointed out that the ' throw\ldquo ; was irregular throughout Groups and , and the effect on the individual experiments is very marked ; nevertheless the values of derived from the fifteen experiments in those groups are in practical agreement with the values deduced from the experiments in Series II .
( 4 .
) Personal errors of observation which , from their very nature , are ullikely to recur at regular intervals or over identical ranges .
These , no doubt , are the origin of many discrepancies , the strain on the observer during these experiments being great .
The cumulative effect of all the above causes of irregularity may at tinues be large .
In Series I. differences of 1 in 500 occasionally present themselves , although , in Series II .
, there is only one case in which the difference between the mean and the curve-numbers exceeds 1 in 1000 , and only two others in which it exceeds 1 in 2000 .
The values of deduced from the different groups ( using the numbers given by respective smooth curves ) are , however , in much closer agreement , the divergence being less than 1 in 4000 ( see Table XLIII It is worth mentioning that in previous years we found the agreement between individual experiments in the same group ( having the same E.M.F. ) was closer than that shown in the above Table ; but , on the other hand , the values of and of water equivalent given by different groups ( or by members of the same group when the number of cells was changed ) , differed hopelessly , in some cases by more than 1 per cent. , showing that constant causes of error had not been eliminated .
attribute the improvement ( 1 ) to the alteration in the form of stirrer , for the mechanical work done in previous years varied capriciously , although no doubt the thermometer bulb was more thoroughly " " washed\ldquo ; than before the alteration ; ( 2 ) to the correction ( now introduced for the first time ) for the difference between the temperature of the wire and that indicated by the thermometer .
As it is more convenient ) examine the values of for integral values of the temperature , we have read off from the smooth curves the values of for every two degrees covered by our range , and also the values at C. and C. These are given in the following table .
TABLE XLI.\mdash ; Values of from Smoothed Curve .
Taking the data given in Columns 4 and 6 of the above table , we give an example of the final calculation of our results .
Using the notation of pp. 478-481 , we the following values THE CHANICAL EQUIVALENT OF HEAT .
volts .
Substituting these values in equation ( 15 ) and dividing , we obtain As the difference in weight of water here amounts to nearly 120 grms. , it is probable that the value obtained in this case is the most reliable .
In the same manner using the values of and given in Columlls 4 and 5 , we get And again from Columus 5 and 6 , Hence mean value of deduced from Series II .
Hence , adoptin C. as the standard temperature , the Specific Heat of Water Also by means of equation ( 15 ) we get the following values of Columns 4 and 6 Mean This value of , as pointed out on , is entirely of the value ssigned to the water equivalent of the lorimeter .
Performing the operations of equation 16 , we find that the water equivalent at * See p. 388 , supra .
If we mean the results from Columns 4 and 5 and 5 and 6 , we must of course obtain the same resuIt as that given by Columns 4 and 6 .
We have however given the numbers in the form as they show the nature of the agreement between the different groups .
Over the range to C. C. in term.s of water at C. grms. ; and from the operations cated in equation ( 17 ) , we get water equivalent at C. in terms of water C. glms .
Hence Water equivalent We are now in a position to find the value of for any weight of water at any temperature , and the value of can then be found by equation ( 18 ) .
The tollowing Table gives the value of for each weight of water temperatures 1 , and C. TABLE XLII.\mdash ; Value of at , and Substituting the values of ( Table XLI .
) and in equation ( 18 ) , we get the following values of J. The first column shows the group of experiments from which each value is derived .
TABLE XLIII.\mdash ; Values of given by each Group at different Temperatures .
We have in the above Table given the values resulting from the calculation diTerent temperatures since the limit of our experimental errors is thus indicated ; for the values of ought ( in the absence of experimental errors ) to identical at all temperatures .
The close agreement between the values from gro.ups , and from the same group at different temperatures , is a satisfactory * Over the range to C. THE MECHANICAL EQUIVALENT OF the accuracy of our determination of the water equivalent of the calorimeter , and the changes in it and in the capacity for heat of the water .
If we reject Group ( and we have already shown that it has little yalue ) the esults are practically identical .
Hence if we assume\mdash ; ( 1 .
) The unit of resistance as defined in the ' B.A. Report , ' 1892 .
( 2 .
) That the E.M.F. of the Cayendish standard Clark cell at C. volts .
* ( 3 .
) That the thermal unitquantity of heat required to raise 1 .
of water through 1o C. at C. The most probable value of This , by reduction , gives the following:\mdash ; kilogrammetres in latitude of Greenwich .
ft.-lbs .
per thermal unit in latitude of Greenwich .
, , , \ldquo ; , SECTION \mdash ; DISCUSSION 0F THE RESULTS .
As stated in the Introduction , we proposed to determine the value of in terms of the thermal unit there defined , , the quantity of heat required to raise unit mass of water through C. at C. ROWLAND has preferred to tabulate his results by giving the changes in the numerical value of caused by changes in the capacity for heat of water .
We can , however , deduce from his table the expression for the change in the specific heat of water over our range .
Expressed in the same form as above , it becomes \mdash ; 15 .
The difference between our results on this point is marked , and it is evident that further investigation is required .
ROWLAND finds that the minimum value lies between and C. ; we hope to carry the ation beyond that temperature , snd some explanation of the difference in our results may then present itself .
The whole question is probably one of , and possibly our of this part of the subject may bring our results into closer agreement , ROWLAND himself ( 1 ) .
198 ) points out that the whole matter depends on a small difference which he * If we assume the ] ] .
of our cells to be the same as that of the Cavendish standard ( and we are inclined to think we have over-estimated the difference ) , we get The value obtained by us in supra , p. 365 .
denotes by the letter .
This quantity ) ears to be of the same nature as which we term in the platinum-thermometer formulae .
quantity had to be mined from a difference of the order of C. between the air- mercury-thermometers .
The method adopted for finding the value of the quantity would appear much less liable to ROWLAND himself was evidently allxious to throw more light on this portion of for he performed , with great care , a series of experiments by the of mixtures .
These are given on pp. 125 to 130 ; he has not , however , tabulated results .
We give , with one exception ( No. 10 ) , the results of all these which did ) involve temperatures aboye C. TABLE XLIV .
Specific Heat of Water by the method of Mixture ( ROWLAND ) .
It is thus apparent that the results of ROWLAND'S investigation by this method in closer agreement with our conclusions than they are with his own .
* We do however , attach much importance to this point , since\mdash ; as he justly method by the mecbanical equivalent is of far greater value .
*Our valuo 000266 .
THE MECHANICAL EQUIVALENT OF The values of deduced from the two groups of Series I. and the middle group of II .
are dependent upon the value of the water equivalent deduced from the and lowest weights of Series II .
The agreement in the results is a sufficient of the accuracy of the value assigned to that equivalent .
The coefficient of increase in the specific heat of the calorimeter may appear large .
TOMLINSON ( ' .
Soc. Proc 1885 ) , gives the following as the specific heat of copper zinc :\mdash ; Specific heat of coppe zinc \ldquo ; , , he quotes the values given by ] as Copper Zinc the mean of these results in the same form as that given to the expresfor the water equivalent , we get Specific heat of copper , zinc .
, TOMLINSON states that when reducing his observations , he used REGNAULT'S expresfor the capacity for heat of water , Unfortunately the account of these experiments given in the ' Proceedings ' does not supply us with sufficient data to estimate the changes in the coefficient of that would be caused by assuming our expression for the specific heat of water .
We are , unable to say how far our rate of increase in the water equivalent agrees with the above determinations .
We are , however , enabled to state that the values of the specific heats at C. would be reduced by about 6 parts in 1000 , on the assumption that our formula holds over so large a range .
Our calorimeter may be regarded as a mixture of copper , zinc , gold , glass , and mercury , and the coefficient of increase of such a mixture can only be obtained by direct expel.iment .
, in ordel to compare his results with those of JOULE , expressed both in kilogrammetres at Baltimore , and gave the numerical value of in terms of water at differing temperatures , assuming his own value of the specific heat of water .
In order to make a similar comparison we give our results in the same form .
The following Table is extracted from the Appendix to ROWLAND'S paper , with the exception of the last column , which we have added .
paper ( Table Ir .
, p. 497 ) contains an misprint in the coefficient of Comparative values of given by JOULE , ROWLAND , and ROWLAND remarks as follows : " " JOULE rejected quite a number of his results , but I have thought it best to include them , giving them small weight however .
In this way we obtain a value of JOULE'S experiments of at C. , my value at this point being ; the difference amounts to 1 in 550 only Adopting this method of comparison we get ( at C. , and expressed in kilogrammetres at Baltimore ) : JOULE .
ROWLA .
GRIFFITHS .
Thus our difference from JOULE amounts to 1 in 350 , and our difference ROWLAND amounts to 1 in The difference between ROWLAND'S determinations of the changes in the specific heat of water and ours , would cause the values of to be identical if expressed terms of a thermal unit at C. , and below that temperature ROWLAND'S value would be the greater .
[ Notes by E. H. G. , added April , 1893 .
1 .
No change in the value of the various units , or constants , involved in *Value deduced by ROWLAND by assuming his own value of the ohm .
If we attach equal weight to the diffel.ent values given by JOULE ( ROWLAND in obtaining the numbers attached arbitrary values to different experiments and mothods ) we obtain mean value at C. , i.e. , at C. This exceeds our value by 1 in 4281 .
If we assume the .
of our cells to be that of the Cavendish standard , our valuo wonld and these differences would become 1 in 386 and 1 in 1260 respectively .
THE MECHANICAL EQUIVALENT HEAT .
investigation could bring our results into absolute agreement with those obtained by ROWLAND , since , owing to the difference in the expressions for the temperature coefficients of the specific heat of water , it is inevitable that our should agree at some one temperature , but must necessarily differ when expressed in terms of a thermal unit at any other temperature , and , , changes in the values of units would only alter the temperature of reement .
For example , Dr. GUILLAUME has pointed out to me that the experiments of Commandant DEFFORGES lead to the conclusion that the value of , at Greenwich , should be increased from .
to 98124 centims. A similar correction would slightly increase value , and thus cause our point of reement to be about C. instead of C. Again , the value of , assumed by Lord RAYLEIGH ( ' Phil. Trans A. , 1884 , p. 427 ) , would have to be slightly increased , and the resulting values of the electro-chemical equiyalert of silver , and of the absolute ectromotive force of a Clark cell , would require modifi- cation , but the only result of any such change would , as before , be to the temperature of agreement .
It is , therefore , evident that the difference is chiefly due to errors in therrnometry .
An error of C. , in the value of the range to C. , would suffice to explain the discrepancy veen the results , and the close agreement in the range value , as obtained from our standard and the Paris standard ( see p. 480 , appears to me to warrant a brief criticism of ROWLAND'S methods and conclusions .
The science of exact thermometry has made great strides during the past fourteen years , and , no doubt , much of this progress is due to ROWLAND'S initiative , for his work undoubtedly marks a distinct advance in this subject .
Great , however , as bis services have been , he would , I feel sure , be the first to admit that his investigations of 1878-79 , by no means exhaust the ities of further progress in the science of thermometry .
In his paper , he constantly icates possible causes of error , and , in some cases , makes valuable suggestions as to methods by which they may be eliminated .
In a previous paper ( Phil. Trans A. , 1891 , p. 155 ) it was pointed out by Mr. CALLENDAR and myself that ' the indications of the air-thermometer cannot be relied on beyond C the limit of of a single reading of the barometer On p. 95 of his paper , ROWLAND gives the data on which the following statements are based .
A difference of millim. in the reading of his barometer gave a difference of C. in the air-temperature , and an error of C. in the imation of the temperature of his barometer-column would produce the same effect .
From his remarks , it is evident that he regarded his possible error from this cause alone C. Again , his temperatures depended on the direct observation of the height of a mercury manometer column by means of a cathetometer , and everyone who has worked with such an instrument knows the difficulty of obtaining surements accurate to millim. ( It is true that small differences can be read to a much closer degree of accuracy by means of a microscope fitted with a micrometer but this method is not applicable when large differences have to be read , nor does appear to have loeen adopted by ROWLAND .
) p. 94 , ROWLAND emarks that repeating the observations upon an accurate scale , they ' agreed with the scale within two or three hundredths of a mi]limetre , which was as near as I could read such an object and millim. appears to have been equivalent to C. the ture of the manometer column was only ascertained by means thermolneter placed alongside it , and the uncertainty thus introduced might double the rescllting error .
A study of RowLAND'S Tables , XI .
to , leads to the conclusion that the crepancy between individual observations , due to the above and other causes , in cases amounted to as much as to C. The following extract will , I think , show that ROWLAND did not rely upon observations to C. In his concluding remarks he says : " " I now believe with the to the .-thermometer of an artificial atmosphere of constant pressure , we could be reasonably certain of obtaining the temperature at any point up to C. , within C. from the mean of two or three observations .
I believe that my own thermometers scarcely differ much more than that from the absolute scale any point up to C. , but they represent the mean of eight tions .
However , there is an uncertainty of C. at the C. point , owing to uncertainty the value of .
But , taking ] , I hardly think that the point is uncertain to more than that amount for the thermometers Nos. 6163 , 6165 , and 6166 The causes of error above enumerated do not affect the platinum determinations to the same extent .
It is true that the values of ( which to RoWLAND'S quantity , ) , are dependent upon the boilin -points of water sulphur , which are again dependent on barometer readings , but , on other an error of as much as C. , at those points , would produce so little change in to leave the value of our range unaltered , although it would slightly affect the elevation ; more especially , it should be noticed that any small changes of which may take place during the of the thermometer with thermometers do not affect the results .
The above considerations appear to me to justify the conclusion that the value of ROWLAND'S temperature range , to C. , may be in error by as much as C. , and , if this is the case , the values of the C.G.S. unit of heat , and of the temperature coefficients of the specific heat of water , as deduced from his experiments , may be identical with ours .
2 .
I have , since the publication of an abstract of this paper , received several munications criticising our statement , that we differ from results obtained by JOULE , by " " 1 part in 350 and my correspondents have out that the ratio , 772.55 to 778.99 , is very different from that of 349 to 350 .
would , therefore , appear necessary that I should indicate , more fully than I THE MECHANICAL QUIVALHiNT OF done , the ground upon which we based the above statement .
JOULE nself , in 1879 , made a careful comparison of his ther1nometer with one which had been standardized by ROWLAND Therm. .
The results were published as an appendix to ROWLAND'S paper in the 'Proceedings of the American Academy ' for March , 1880 .
A considerable correction in the values obtained by JOULE was then found to be necessary .
Again , JOULE had assumed REGNAULT'S expression for the specific heat of water , and a correction for the capacity for heat of the calorimeter was also rendered necessary by the changes in the thermonletry .
As the appendix above referred to appears to be but little known in this country , I give one example ( p. 44 , ibid. ) to indicate the nature of the various corrections .
JOULE'S result in 1878 ( friction of water ) Correction for thermometel ' , , capacity for heat . . .
2 , , latitude ( to Baltimore ) .
9 , , vacuum . . . .
\mdash ; 9 Corrected at C. As these conectionoe were published in 1880 , it appears probable that , had they not met with JOULE'S approval , he would have signified his dissent , but I can find no record of any such expression of opinion on his part .
As previously stated , ROWLAND attached what appear to me to be curiously arbitrary values to certain of JOULE'S experiments .
If we acept his verdict as to their tive merits , the mean of the selected experiments , as corrected by ROWLAND ( but reduced by me to temperature , and to latitude Greenwich ) , is -lbs .
; but if we assign equal value to all JOULE'S published results , the corrected mean becomes ft.-lbs .
, and it was on these numbers that our statement was based .
It is extraordinary that , although these results ( or rather the corrections by which they are deduced ) were published as far back as 1880 , we still find that our modern books of reference continue to give 772.55 as the number resulting from JOULE'S own experiments .
] APPENDIX I. of the Null Point\ldquo ; and Method of calculating Results .
As stated in the introduction we append a full the null point i.e. , point at which the radiation is self-eliminated .
Using a similar notation to that on p. 478 , we have as the equation of condition ( 1 ) , where the temperatures are measured from that of the surrounditJg envelope .
For the of simplicity we can assume that the values of and remain constant .
Integrating and putting , and determining the constant from the fact that when , we obtain the equation ( 2 ) .
If there had been no radiation , , and the equation of condition would have been Integrating , and using the same constant as before . . . . . .
( 3 ) .
If we find the points of intersection of ( 2 ) and ( 3 ) one point is that at which the experiment commenced , the other is the point on ( 2 ) at which the radiation is eliminated .
It is more convenient , for experimental work , to obtain an expression involving rather than ; substituting therefore the value of given by ( 3 ) in ( 2 ) , we obtain This equation can be solved for when the values of are known .
In order to obtain , we can take the following observations : Commence an experiment at , note the time when , and again note the time , when Then equation ( 2 ) gives the relations THE CHANICAL EQUIyALENT OF HEAT .
( 5 ) , ( 6 ) From this pair of equations we can eliminate and obtain which is the equation from which to determine So far as we are aware , the solution of these equations cannot be expressed as ) algebraical expansion , except on the assumption that are small .
Making this assumption , and writing , successive approximation leads to ( 8 ) .
The method which we have found best adapted to the numerical solution of such equations amounts practically to tracing their graph , using , for , in equation ( 7 ) , and as abscissa and ordinate .
A rough approximation to the value of is soon obtained , and afterwards by means of a table of logarithms as close an approximation as is necessary can be found .
Since does not occur as an exponential , its value is at once found from ( 5 ) or ( 6 ) .
Equation ( 4 ) can be solved for in a similar manner to equation ( 7 ) , usin and as abscissa and ordinate .
A rough approximation to the value of is given by equation ( 8 ) , .
The forms given to equations ( 4 ) and ( 7 ) are those which we have found most convenient from which to calculate the values off ( ) and .
Besides being thus able to eliminate radiation , we can at the same time eliminate the change in all those quantities whose rate of variation can be expressed as a linear function of the temperature .
Thus , on p. 477 , we have shown that when the variations in and are considered , their fulJ values are ( 6 ) shows that is not altogether at our disposal , but must be less than .
For , the calorimeter would reach its final state , where the loss by radiation balances supply of could never attain the temperature and , hence , equation ( 1 ) becomes .
( 9 ) By neglecting squares and products of small quantities ( 9 ) can be written in the same form as ( 1 ) , thus , and the values of and become ' .
( 10 ) , where Let denote the values of and when a weight of water and an E.M.F. of Clark cells are used , the first suffix referring to the weight , the second to the number of Clark cells ; further , as before , let be the water equivalent .
Then ( 11 ) ( is written for as its value is dependent on ) .
Similarly ( 12 ) .
Hence , by division of ( 11 ) and ( 12 ) , ( 13 ) .
Equation ( 13 ) gives the means of calculating the water equivalent .
Writing for and respectively , we have ( 14 ) , which , with ( 1 gives ( 15 ) .
*This does not introduce , the heat developed by the stirrer , but this can be found by experiments and due allowance made for it .
THE CHANICAL E OF HEAT .
If , the temperature coefficient of the resistance of the wire , is determined by preliminary observations , and we assume where is a constant ( p. 400 ) , we have with ( 15 ) sufficient data to determine the values of .
Equation ( 11 ) then gives . . . .
( 16 ) .
Again equation ( 10 ) gives ( 17 ) , and ( 18 ) , which are two equations to give and Similarly , by performing the operations and , we could determine the value of ; and since ( p. 480 ) and we can find both and , the coefficients of increase in specific heat of water and of the calorimeter .
APPENDIX II .
The Regulator intaining the at Temperature .
A narrow glass feet in length , was fixed in a horizontal position on one of the walls .
This tube contained chloride of silver , and a stream of dry ammonia gas at a low temperature was passed through it until the compound was formed .
This compound , as pointed out by ISAMBERT , dissociates at ordinary temperatures , the pressure of the vapour at about C. altering by more than 12 millims. per degree .
Care had to be taken to completely saturate the , for the pressure of the vapour changes at ordinary temperatures by a much smaller amount .
The horizontal tube communicated with a gas regulator of the ordinary except that the diameter of the regulator tube could be made of any size , since we are dealing with the pressure of a saturated vapour , not with that of a gas , and thus the changes in volume caused by a movement of the mercury column could be disregarded , the diameter of the smallest pipe in our * The exterior of this tube was blackened .
This substance was suggested by Mr. C. T. HEYCOC regulator exceeding 1 centim. In order to diminish the effect of changes in atmospheric pressure , some air was allowed to remain in the horizontal tube .
As column of mercury communicating with the regulator was about 800 millims. height , this residual air was under a of two , and , its vohlm being very small , and the diameter of the mercury tube large , a small movemen sufficed to compensate changes in atmospheric pressure .
Thus , an alteration 20 millims. in the barometer produced a change of only a fraction of a degree in resulting laboratory temperature .
The gas , after passing through the communicated with a large Doulton ware , in which small pilot lights always kept burning by means of an independent supply .
An entrance and an tube , closed by glass taps , were so fixed in the vertical tube containing the that , by their means , the instrument could be adjusted .
The action of the apparatus is , in many ways , interesting .
On a day when the would be , if the room was ccupied , the presence of one person in the room would diminish the supply , whereas presence of two or three would cause it to be entirely cut off .
We were surprised to learn , by this means , the amount of heat to a room by the presence of a human being .

