]\gt ; Conduction of heat place through thin layers of the liquid , and thus in interpreting the results , the conditions at the surfaces ) ltlng the liquid layer are of primary importance .
It has been assumed by each observer that contiguous sul'faces of any two media are in all circumstances at the same temperature .
This , however , is contradicted by some anthorities , so it would seem important to have independent rosults based on experiments in which the liquid layer is of considerable thickness .
It should also be noted that in methods employing thin layers the temperature varies so apidly in passing from one } face to the other that the liquid forms a by no means very homogeneous medium .
This is the more important because experiments indicate that the conductivity of most if not all liquids increases rapidly as rises .
The carl.ied out in the atory at the suggestion of Professor J. J. Thomson , to whom I am much indebted for estious as to the form of the tipparatuh and the methods to be empJoyed .
series of experiments were made with different apparatus .
the earlier series it was found that the appal.atus was too to conveniently worked , and few results of a satisfactory nature obtained .
In the second series the appal.atus was much educed i size , though otherwise closely esembling that first employed .
It will thus be sufficient to describs the second form and supply data as to size of the first .
The liquid was contained in a -ooden tub sides , cm .
in diameter , which was carefully fitted up by the at Cavendish Laboratory .
Not far below the and at equal distances apart were fixed three conical wooden pegs .
The axes of the pegs formed parts of radii of a horizontal section of the tub , projecting inwards from the surface to a distance what exceeding 2 cm .
The pegs supported a ilat , thin-bottomed dish of tin-plate , cm .
in diameter , whose base was thus horizontal .
The liquid was poured into the tub till it reached base of the dish .
The liquid surface strictly horizontal , it was easy to judge by the eye whether the dish so also ; if not the tub had to be adjusted till it was so .
It was then advisable to stir the liquid to make sure that no air bubbles remained clinging to the dish .
The bottom of the dish was about cm .
above that of the tub .
The method required the temperature to be measured at a known depth below the liquid sulface .
This end was secured by the electrical resistance of a fine straight platinum wire cm .
long , which was supported at a depth of cm .
below the surface by two small trestles of glass .
These were fixed in the bottom of the tub and projected upwards .
The platinum wire was drawn tight over each end being fused to a much thiokcr picce of copper vire , so that .
C. Chree .
the junctions were exactly at the same depth and immediately below the level of the platinum wire .
Ths copper wires were tied with silk to the vertical parts of the trestles , and , passing straight down , were led throu the bottom of the tub .
The middle point of the platinum wire was vertically below the centre of the dish , and thus its ends were much nearer the axis of the tub than was the rim of the dish .
The strictly position of the wire was tested by poul'ing in water first to the level of the wire and then to that of the dish .
It was thus made certain that when the tub was placed so that the dish was horizontal , the platinum wire was so also .
For most liquids the pegs , wires , \amp ; c. , were secured by sulphur , but for bisulphide of cal'bon this was replaced by asbestos .
During the experiments the tub was placed inside a double-walled wooden box , the space between the walls being stuffed with packing .
The box was provided with a double lid similarly stufted .
To the ends of the copper wires were attached binding screws , bhe wires leading from which passed through cut in the rim of the box .
Thus the wires were in no way disturbed in moving the lid .
This was a point of some , as even the small variations in the electrical resistance prodnced by slightly disturbing the binding screws was apt to affect the accuracy of the observations .
The platinum and its connecting wires formed one of the resistances of a bridge arrangement .
One of the others was a fixed resistance , and the remaining wo were supplied by a wire bridge with a sliding-piece .
The resistance of the platinum wire varies with the temperature , and\mdash ; at least for smallvariations\mdash ; its change is propor- tional to the of temperature .
The current was supplied by a single element\mdash ; a small Daniell or Leclanche .
A galvanometer , whose resistance could be redllced to of an ohm , measured the variation from a balance betweeu the resistances .
The usual precautions in dealing with small resistances had to be taken ; in particular it was found difficult to avoid producing thermoelectric currents if the sliding-piece were moved .
In the first apparatus the tub was cm .
in diameter , and dish .
The platinum was coiled in a spiral round a fine horizontal glass tube at a mean depth of cm .
below the liquid surface and about 7 cm .
above the bottom of the tub .
The length of the spiral was less than the radius of the tub .
The method of conducting the experiment was as follows : tub was put inside the box and filled with the liquid to the level of the dish .
The box having been adjusted till the dish was horizontal , the lid was put on .
The sliding-piece of the bridge wire was then moved till no culTent traversed the galvanometer .
As the tnb and liquid were in general at slightly different temperatures to begin wirh , some time elapsed before the galvanometer reading became constant .
Conduction of in Liquids .
31 ' ; When this had occurred the sliding-piece was again moved till there was no current through the galvanomsber .
It was then unnecessary to move the sliding-piece again , unless the deflection became greater than was usual in the experiment .
When the galvanometer had remained some time constant , the lid of the box was removed and some hot water rapidly poured into ths dish , care being taken that none splashed over into the tub .
Sometimes the lid was immediately eplaced and left on during the whole of the experiment ; on other occasions the } was after a certain interval removed by a siphon , ready filled for the purpose .
This always lefb a small quantity of water sllfficient to cover the base of the dish without into drops .
The battery w.as connected with a key , and there was another in galvanometer circuit .
In the earlier experiments these were depressed in cIose succession at intervals of one minute , and the conscquent kick or deflection of the needle observed .
Subsequently it was found more convenient to uss a constant battery , and to keep both keys down during the whole course of the experiment and some time eviously .
Both methods were employed for most of the liquids examined , and no difference was detected in the results .
The temperature of the platinum wire was seldom raised as much as C. during the experiment , and consequently the disturbance of the balance in the Wheatstome 's bridge was small .
Thus the current through the galvanometer could be taken as directly proportional to the in the resistance of bhe platinum , and so to the rise in its temperature .
Immediately subseqaent to the application of the hot water there was a decided increase of the galvamometer reading which ceased very shortly .
The reading then remained almost for minutes .
It then began to increase rapidly and continued to rise for a considerable time , though the rate of change comparatively soon to decrease .
At first with the larger tub it was attempted to determine the interval that elapsed before the ceased to increase .
This was , however , found impracbicable , as it required several hours to reach this epoch ; and after an hour and a half the rate of hang was so slow that the least variation in the temperature of the laboratory was sutiicient to upset the Even with the smaller apparatus this was not a quantity to be conveniently observed .
It was found much easier to mine the much shorter interval that elapsed before the platinum wire was being most rapidly heated .
This interval could also be expressed conveniently by means of the mathematical theory in terms of the conductivity and other properbies of the liquid , and so its determination was sufficient for the purpose in vie The galvanometer could be made so sensitive that with a singls VOL. XLIII .
Mr. C. Chree .
small cell a deflection exceeding 300 divisions of a miIlimetre scale could be obtained for a rise of one degree in the temperature of the platinum .
When so sensitive as this , however , the ometer was too much exposed to the disturbing influences of adjacent currents or the movement of magnets in neighbouring rooms .
From eighty to a hundred scale divisions to a degree usually gave the best results , and had the advantage of keeping the spot of light near the oentre of the scale during the whole experlment , without any movement of the sliding-piece .
The sensitiveness was most easily determined by finding how far the sliding-piece on the bridge had to be moved , when the battery was on , to produce a given change in the galvanometer reading .
This test was usually applied at the beginning and end of each experiment , as a change in the sensitiveness during the observations might lead to erroneous conclusions .
slight displacement of the controlling magnet of { he galvanometer may occur without aflecting to any noticeable extent the position of the , and so without some such test as the above a change in the sensitiveness might escape detection .
Theory .
Let denote the temperature , the density , the specific heat , and the conductivity of a given liquid .
Suppose the liquid to extend to infinity in every direction , and over the entire plane a uniform supply of heat to be distributed at a rate given at the time counted from the first application of the heat , by the function per ullit area ; then at the ance x from the plane of application the temperature at time is given by .
( 1 ) In the present experiment the base of the dish ) to the plane , and is to be regarded as proportional to the rate at which heat is conveyed from the dish to the liquid .
It is true of course that the liquid exists ] on one side of the plane , and does not extend to infinity in any direction .
Doubtless the base of the tub tends to reflect the heat that has passed downwards through the liquid , but in the apparatus actually used any such reflected heat vould be extremely small , and only the most trifing part of even this effect would show itself withiu the time of the experiment .
Since the of the platinum wire was much less than the diameter of the dish , which was in turn considerably less than that of the tub , the limitation in the horizontal direction would appear of small conse quence .
In fact when a larger quantity of hot water thau was usually employed was poured into the dish , a delicate thermometer indicated Conduction of Beat in Liquids .
.3.5 a temperature in the liquid that was sensibly constant at a constant depth except close to the sides of the tub .
The abseuce.of liquid on the oegative side of the plane might appear a radical defect .
It is clear , however , that in the supposed infinite liquid will pass into the liquid on each side of this plane , and the existence of the liquid on the one side merely ensures that is the precise amount passing into the liquid on the other side .
But the law of diffusion on either side of the plane can depend only on the heat supplied to that plane , and must be independent of the precise nism by which the ] is regulated .
For onr present purpose it is sufficient to know that is proportional to the rate at which heat passes into the liquid from the dish , which may be determined by a double observation as follows .
The tub bein filled with liquid up to the level of the disl ] , a certain quantity of water heated to a definite temperature is suddenly pou into the dish .
By means of a } } atch , and a delicate thermometer , raised initially to the temperature of the heated water and with it transferred to the dish , the law of cooling of the water is determined .
The quantity of heat lost by the dish per unit of time at any required temperature can be easily deduced .
If now the dish bs placed on a non-conducting material , and the law of cooling be obsel.ved when the other circumstances are the same as before , the quantity of heat which leaves the dish per unit time in the first experiment without passing into the tub is at once obtained for the whole ran ge of temperature .
From these two experiments it is not difficult to calcuIate the amount of heat passing into the liquid in the tnb at every instant in that fo of the experiment in which the water poured into the dish was left there .
When a siphon was employed the capacity for heat of the water left in the dish and the dish itself was so small that the heat subsequently transferred to the liquid was negligible .
To a clear understanding of the use of ( 1 ) some knowledge of the expression is desirable .
This is proportional to the temperature existing at a depth in an infinite liquid , originalIy at zero temperature , at a time subsequent to the applioation over entire plane of a unit of heat per unit of area .
The first and second differential coefficients of the above expression are re , and Thus the temperature at depth , counted from the plane gradnally commences to rise and continues to do so for a time Mr. C. Chree .
, after which it steadily falls .
The times at which the increase and decrease are fastest are respectively the smaller and the greater root of the quadratic ; .
and are approximately .
' and Supposing it were ssible suddenly to supply a quantity of heat to the surface of the liquid in the tub and to ensure that no commensurable quantity was subssquently gained or Iost , an observation of the time at which the tcmperatnre at a given depth was fastest , or was stationary , would enable to be determiued at once .
In the actual case the problem is more complex as , though ishing rapidly as increases , is different from zero ; the principle however is practically unchanged .
By differentiation we obtain from terms at the limits . . . . .
Now when is moderately large the terms the limits may be neglected .
This follows from a consideration either of the mathematical form or the physical meaning of those terms .
They are proportional , one to the temperature instantaneously produced at the depth at the time by heat at that instant passing into the liquid from the dish , and the other to the rate of change of this insfantaneous effect .
Now even when heat is being very rapidly communicated to the liquid , as at the commencement of the experiment , the rise in temperature due to conduction at a moderate depth is for a minute or two insignificant , .
Thus when the heat is being communicated very slowly , as is the case at the time at which we shall employ ( 5 ) , terms at the limits are for all practical purposes negligible .
When the temperature of the liquid at depth is rising most rapidly , is a maximum , and so .
From the above reasoning it follows that the time in question must satisfy the equation\mdash ; .
( 6 ) Conduction of Heat in Liquids .
This equation cannot be exactly solved , but an approximate soIution of sufficient accuracy can be obtained .
This gives as a function of , and ; but is determined from the galvanometer and , and can be otherwise determined , thus is at once obtainable .
With the smaller apparatus , when the dish remained unemptied , the vaIue of , when water was in the tub , exceeded ten minutes , and for nearly all other liquids it is greater .
The can be eplaced by the summation\mdash ; ( 7 ) is proportional to the heat transmitted to the liquid during the interval 7 , and is the time between the middle of interval and the epoch of swiftest rise of temperature .
It is not necessary to take the same throughout ; thus the of the experiment when varies rapidly , must be taken smaller than subsequently .
The terms in the summation answering to the few minutes of the experiment are always very small .
When the water was siphoned from the dish , any gain or loss of heat through the dish subsequent to the operation was very small compared to that given up the liquid previously .
Thus no serious error will be introduced by supposing after the siphoning .
It will be observed that what the galvanometer readings give is the time when the platinum wire is heating fastest , while the equation gives this epoch for the liquid at the same depth as the wire .
Since the temperatures of the media are changing very slowly , it is scarcely conceivable that they could differ by a fnite quantity , or that their rates of change should not be practically alike .
The as sumption made in the present method is of a totally ferent o from that made by previous observers dealing with thin layers of liquid .
Their assumptions would be equalled only by the dish and the liquid touching it to be always identical in temperature .
the absolute qua1ltity of heat initially given to the dish is of no importance , except in so .
as it modifies the rate at which heat is subsequently communicated to the liquid .
Experimelltall was found that both the and the temperatul.e of the water poured into the dish could be varied to a considerable extent without sensibly altering the epoch of quickest rise of temperature .
When the was siphoned the initial quantity of heat was of still less importance .
With most liquids , however , the water was heated to a fixed temperature , C. , and a meaf ; urod Mr. C. Chree .
quantity was ponred into the dish .
With bisulphide of carbon of conrse a much lower temperature was employed , but even then its heated top layer evaporated so fast as to aflect the contact of the liquid with the dish .
For these reasons in all experiments on the bisulphide the water was siphoned out of the disb .
In the earlier part of each experiment the heat was of course concentrated chiefly in the upper layers .
Still as about an inch intervened between the dish and the platinum , the variation of tempera .
ture in by far the greater portion of the liquid layerwas comparatively small .
Thus the error due to treating the conductivity as independent of the temperature caunot be great .
There are two possible disturbing agencies which require comment .
Any difference of temperature between the two copper-platinum junctions in the liquid would produce a thermoelectric current .
Care was taken , however , that the junctions should be as nearly as possible in the same horizontal plane .
At tlJe depth in question the greatest possible difference between the temperatures at two points differing a few millimetres in depth could not exceed a small fraction of one degree .
Thus the thermoelectric current , if existing , must have been very small , and necessarily its variation , from which alone any error could alise , must have been very trifling .
urther , the neutral point of copper and ] atinum is only about , so they would under the circumstances form a very weak couple .
An attempt was in fact made to employ a thermoelectric couple of iron and lead , whose neutral point is over , one junction being in the liquid and the other maintained at a constant temperature .
This , however , failed completely , owing to want of sensitiveness .
Thus there are various priori grounds for neglecting the thermoeIectric effect in the actual experiment .
This view was further justified by actual trial , first by finding the deflection that followed when one of the junctions was suddenly heated to a considerable temperature , second by cutting out the battery during the usual experiment , and whether shunting the platinum wire affected the galvano .
Finally , in the various experimemts on any one liquid the battery current traversed the platinum wire sometimes in one direction sometimes in the other ; and thus any possible thermoelectric effect must have tended sometimes to increase and sometimes to diminish the rate of variation of the galvanometer reading .
The small iation in the observed times of most rapid variation is thus sufficient proof of the small disturbing action of the thermoelectric effect , and the variation in the direction of the battery current would further tend to eliminate any such small effect if existent .
Another disturbing cause existed in the case of the sulphuric acid soIutions .
These conducted ectrolytiCally , and also attacked the copper wirss .
By covering these wires with shellac varnish this was Conduction of Heat in Liquids .
prevented .
With the weaker soIutions a slight coating was sufficient , but the stronger it had to be frequently renewed .
In the cooling experiments , if we suppose the mass of water in the dish to remain constant , the rate at which the thermometer falls at any iustant is directly proportional to the rate at which heat is leaving the dish .
this heat some passes into the material below the dish , and some is lost by radiation from the water and the sides of the dish .
The material om which the cooling was slowest was packing , and the heat given up to it seemed very small .
Of by far the greater part occurred in the first minute , and this is preclsely the time when a small error in calculating the heat given to the liquid in the experiments on conductivity is of least importance .
Thus in default of more accurate knowledge , the loss of heat by the dish when on packing was taken as representing the loss by radiation when on a liquid .
Since the absolute amount of heat given to the liquid is not required , but only the ratio of { he quantities given up for each minute or half minute of the experiment , absolutely no error would be introduced by neglecting the conductivity of the packing , proided the heat passing into the packing followed the same law as that into the liquid .
The liquids whose conductivities were determined are water , sulphuric acid solutions of various strengths , bisulphide of carbon , one solution of methylated spirit , paraffin , and turpenbine oils .
For water , methylated spirit , and paraffin two series of observations were made , the water being siphoned from the dish in one case , and left in the other .
In the case of turpentine no observations were made with the water siphoned .
For the ] phuric acid soIutions and the bisulphide of carbon the siphon was always used .
It was found that the conductivity and the rate of cooling of the dish were nearly independent of the strength of the sulphuric acid solution , and diflered little the corresponding quantities for water .
The law of cooling on bisulphide of carbon also closely resembled that on water .
The reIative co1lductivities of these liquids would thus in all probability be most correctly obtained by referring them to the value obtained water by the method employing the siphon .
The following table gives the quantity of heat given up to the liquids in consecutive minutes or balf minutos of the experiment , so far as is required in calculating the ductivity .
The unit employed is arbitrary , but is the same throughout : \mdash ; The times required for the dish to cool from to on water and on the sulphuric acid solution were almost identical .
Excluding the first half minute , the cooling on the solution was slightly but decidedly faster for the first half of the period .
In the bis[llphids of carbon the experiment was made at a much lower temperature , which accounts for the comparatively small quantity of heat given to the liquid in the third half minute .
In fact , considering the small temperature excess the rate at which the dish lost heat on the bisulphide was initially extremely rapid .
On turpentine for the first few minutes the Ioss of heat was much faster than on packing ; the rates then began to approach , and after the first nine minutes could not with certainty be said to differ .
This coincidence lasted for the nexb five minutes or more , during which the observations were continued .
Excluding the first minute , the cooWing on turpentine was decidedly slower than on any other liquid .
is due to the low conducting power and small specific heat of the liquid , in virtue of which the top layer soon became a sort of barrier to the penetration of the heat .
table gives the dsnsity , and speoifio heat , , of the liquids , and the time , , in minutes , after the heating commenced before the temperature of the platinum wire was rising fastest:\mdash ; Conduction of Heat in Liquids .
In each case the density was taken with a hydrometer at or near the temperature of the iment , and the specific heats are taken rlirectIy from ' Watts ' Dictionary of Chemistry , ' or obtained by polation from tables given there .
In the case of the last three liquids the specific heat was also obtained experimentally , as these liquids vary somewhat in composition .
The results so obtained were somewhat rough , but were sufficiently good to act as a check on the values taken .
Small errors in the specific , heat are not of importance , as the probable error in the experiments on the conductivity amounts to at least several per cent. of the numbers taken .
Furthel .
, the mean temperature to which the conductivity should be refelTed is also a somewhat doubtful matter .
The introduction of small impurities in the liquids did llot appreciably alter .
For instance , small quantities of salt were put into the methylated spirit , and small quantities of the latter into water without producing any apparent effect .
It would thus appear that the absolute purity of the liquids used is not of much consequence .
Care was , ever , taken to keep them as pure as possible , the bisulphide of carbon in particular being redistilled before use .
In every case the value is the mean of a good many experiments , and , as a rule , the individual experiments well together .
Thus , when the siphon was the values obtained for varied from to 11 for water , from to 14 for methylated spirit , and from to 15 for turpentine .
When the siphon was used the extreme in the numbers obtained for were about as large as in the other method , and thus the agreement between the experiments was really not quite so good .
This was only to be expected , as there was ecessarily some slight variation in the time taken to siphon and in the result of the operation .
Iu obtaining a mean value for the following method was adopted : Mr. C. Chree .
\mdash ; A table was formed giving the increase in the galvanometer readings for each minute of each experiment .
If there had been much variation in the sensitiveness of the galvanometer , the numbers obtained from each experiment were multiplisd by a number varying inversely as the total increase in the readings during the fifteen minutes of that experiment subsequent to the ] ication of the heat .
The numbers for each minute were then added together , and the sum gave the mean rate of ] ] eating for the minute in question .
From these rates the time of fastest heating could be easily calculated , or could be oblained graphically by constructing ths curve whose abscissae were the times elapsed since the heating , and whose ordinates were proportional to the rates of heating .
It will be best to consider first the experiments in which the water was siphoned from the disb , as the arithmetic required to obtain the conductivity from the equation ( 7 ) is then comparatively simple .
It is assumed that the heat passed into the liquid for the first three halfminutes according to the law given in Table I , and that subsequently no heat at all was either given or lost through the dish .
As it took some time to perform the siphoning , .
and there were no doubt slight variations in the small quantity of water left in the dish , the above is only approximately true , but the multiplication of the observations would tend to eliminate the errors .
If be denoted by X , then from ( 7 ) and Table I , since we have for water\mdash ; From this equation X must be obtained by triaL If stand for the left-hand side of the equation , it will be found that the corresponding values of X and are as follows:\mdash ; The valuo of is best found by considering the logarithm of the several lines in succession .
The following are the values of lines : \mdash ; Conduction of Heat in xids .
From the values of a very close approximation to X may be obtained by heorem , which gives .
This is a much eloser degree of approximation than is at all necessary , considering the possible size of the experimental errors .
Since depth of wire cm .
, get finally , the units being centimetre and minute , This corresponds to a temperature of about C. The law of cooling of the dish on sulphuric acid solutions so nearly resembled that on water , that it will be sufficient to take the results given in Table I for a solution of density , and combine them with the corresponding results for water in a ratio proportionate to the strength of the intermediate solution .
We thus obtain as proportional to the heat given to the liquid in the first three half-minutes the following values : Employing and X in the same ense as for water , it wilf be found , precisely as in the previous case , that the following results are true for the various solutions .
The last column gives the temperature to which the conductivity belongs . .
C. Chree .
Neither the method nor the theory is so extremely accurats that any value can be attached to the third significant figure in the value for in assignlng absolute values for the conductivity .
In assigning relative values , the third figure would have some weight in liquids in which the heat was applied so similarly as in the case of water and the above solutions .
Since , however , the temperatures of the experiments were not identica ] , and the conductivity unquestionably inwith the temperature , it would probably be unsafe to deduce the above numbers any more precise conclusion than that the presence of a very considerable quantity of sulphuric acid produces an emely small change in the conductivity for heat of water .
It might also be c , onsidered almost certain that the time at which the temperature was rising fastest diminished as the density of the solution increased .
This signifies that the velocity with which heatwaves travel , or the temperature Weber , is greater the stronger the soIution .
The liqnid in the tub in these experiments general stirred up fifteen minutes after the heat had been applied .
It is pretty obvious that the ratio of the galvanometer reading after the stirring to that before should diminish as the temperature conductivity increases .
The ratios so obtained for water and the above solutions were in order , and No very great accuracy can be claimed for these numbers as the determination was somewhat rough , but as independent of the truth of the above statement as the temperature conductivity they are of considerable weight .
'For bisulphide of carl ) , methylated spirit , and paraffin oil , the values.from Tables I and substituted in equation ( 7 ) lead to the esuIts : There sbill remain to be considered the experiments in which the water was not siphoned from the dish .
As an example of apConduction of Heat in Liquids . .
In the case of water , there is a decided though not very serious discrepancy .
The difference in the mean temperature of the two experiments could account for only a small part of this .
The experiments on water were the earliest in which the siphon was used , and the operation took slightly longer and its results were not quite so uniform as in later experiments .
rther , when no siphonmg took pJace , the heat passing into the liquid at the end of the experiment was much larger in the case of water than for the other liquids , and the terms at the limits in ( 5 ) would thus be of slightly greater importance for water for the others .
Also an error of given amount in the experimental determination of the time of most rapid heating would produce the effect the shorter the time , and would thus modify the results for water more than for any other of the liquids , cept bisulphide of carbon .
Thus it was only to be expected that the greatest discrepancy between the results of the two methods should occur in water .
With the larger apparatus results were obtained for water and ylated spirit , of the same as in the experiments already described , which , though not pretending to great accuracy , may be of interest as independent evidence of the correctness in the main of the theory .
For the intervals in minutes that elapsed after the application of the heat before the temperature of the wire was rising fastest , the mean of several experiments gave for water , for the spirit .
The water was left undisturbed in the dish , and no accurate observations of the rate of cooling were made .
It was noticed , however , that the dish parted rapidly with its heat , and being only htly deeper than the dish in the small apparatus , it is pretty clear that by far the greater part of the heat was given to the liquid in the tub in the first few minutes .
Thus the experiment would be pretty much akin to the when the water was siphoned in the smaller appara .
Though norant of the law of cooling , we can thus obtain an inferior limit to the conductivity , of a moderate ] close , by the heat to have been instantaneously commuuicated .
This gives from expression ( 3 ) , ) , for water and for methylated spirit , corresponding to temperatures of about C. These as being essentially inferior limits , agree fairly with those of the smaller apparatus .
On the whole , the results of this series of experiments resemble those by Herr Weber .
* The values obtained for the conductivity of watel agree fairly well with his .
The smaller value obtained by Weber for bisulphide of carbon , , would be partly aocounted for by the very considerably lower of his iment .
As this liquid boils at a very low temperature , the * ' Wiedemann , Annalen , ' vol. , pp. 103 , 804 , 472 ; see specially tablo on p. Mr. G. F. Dowdeswel ] .
rate of variation with the temperature of its thermal conducbivity is very probably much above the avera To reduce the resuIts of the present paper to the C.G.S. system of units , it is only necessary to divide them by 60 .
On Rabies By G. F. SWELL , NI .
A. , F.L.S. , F.C.S. Commumcated by Professor VICTOR HoRSLEY , .
Received May 9 , \mdash ; Read June , 1887 .
Numerous as are the communications upon the subject of rables , the paucity of experimental investigation is remarkable ; the disease has remained for upwards of 2000 years , since the first reoorded mention of it by Aristotle , exceedingly obscure in many essential points .
The unparalleled and variable length of its incnbation period has offered the greatest obstacle to systematic examination ; in bhe words of John Hunter in the last century , " " It has defied alike scientific investigation as to its intimate nature , and all remedial measures for its successfnl reatment Lately , however , the results announced to have been attained by M. Pasteur , have promised to remove these obstacIes , and encouraged research by new methods and with fresh visws .
This investigation was commenced early in 1885 , during the prevalence of rabies in and around London .
Two well-marked cases in dogs were obtained , and inoculations with their saliva , taken both during life and shortly after death , were made into the subcutaneous tissue of other animals , but tailed to produce infection .
At that time I was not sufficiently conversant with the resuIts of M. Pasteur 's investigations to place reliance upon his methods of nial inoculation with the cerebro-spinal snbstance of a rabid animal , and I must admit that his statements seemed to me to be im-

