Electromotive Force of different Forms of the Clark Cell .
On the Variation of the Electromotive Force of different Forms of the Clark Standard Cell with Temperature and with Strength of Solution .
" By H. L. Callendar , M.A. , F.R.S. , McDonald Professor of Physics , and H. T. Barnes , M.A.Sc .
, Demonstrator of Physics , McG-ill University , Montreal .
Received August 12 , L897 .
S 1 .
Objects of the Investigation .
The primary object of the present series of experiments was that of equipping the McDonald Physics Building of McGill University with a reliable and accurate set of standard cells , and not that of forming the subject of a communication to any scientific paper .
In the course of the work , however , several points have come under our notice , which we venture to think may be of interest to others engaged in any investigation requiring the employment or construction of such standards .
Among other points , we have devoted special attention to the accurate determination of the temperature-coefficients of various forms of Clark cell ; to the construction of cells free from " diffusion-lag " consequent upon change of temperature ; and to the investigation of the limits of accuracy attainable with Clark cells under both constant and varying temperature conditions .
We have succeeded in making a very simple modification in the Board of Trade form of Clark cell , which makes it equal to any other form in respect of freedom from diffusion-lag ; and we have made several forms of cell hermetically sealed with glass and platinum , which we hope will stand the test of time better than those sealed with wax and marine glue .
We have also made a special investigation of the effect of changes of strength of the solution of zinc sulphate on the E.M.F. , involving determinations of the solubility of zinc sulphate and of the density of the solutions , which appear to lead to very simple formulae of some theoretical interest .
S 2 .
Preliminary Work .
At the outset of our work , we were extremely fortunate in finding the laboratory already equipped by the liberality of Mr. W. C. McDonald , the donor of the Physics Building , with a very fine and complete set of resistance standards and electrical measuring instruments , collected by Professor John Cox , under whose supervision the laboratory was planned and erected .
There were in the collection several portable Clark cells by Muirhead , and a set of Carhart cells by Queen and Co. were shortly added .
k 2 118 Prof. H. L. Callendar and Mr. H. T. Barnes .
Several comparisons of these portable cells were made early in 1894 at different dates and under ordinary laboratory conditions .
The results , when corrected for temperature by means of the enclosed thermometers , showed irregular differences , often amounting to neai lie 3 millivolts .
Of a pair of cells enclosed in one brass case together with a thermometer , one would be sometimes higher and sometimes lower than the other , although the utmost care was observed in using them , and they were both necessarily always under closely similar conditions as regards change of temperature .
The set of Carhart cells were much better in this respect .
They rarely showed irregular differences of more than half a millivolt ; but one of the cells , owing to defective sealing , was generally some 2 millivolts lower than the others , and has since that time fallen still further .
It appeared probable at first that these discrepancies were partly due to inequalities of temperature between the cells and their attached thermometers , and , if so , that they would be inseparable from portable cells of such a form under these conditions. .
About the same time , a number of Clark cells were set up by Professor Callendar and some of the advanced students and demonstrators , in accordance with the form described in the Board of Trade Memorandum , as figured in Glazebrook and Shaw 's ' Practical Physics , ' p. 576 .
Every detail of manipulation and construction was carefully followed , except that the cells were set up in test-tubes-6 inches long , to permit of their being more deeply immersed in water , and that the glass tubes containing the electrodes were provided with mercury cups at the top to facilitate the making and changing of connections .
The cells were kept immersed in water in large glass bottles provided with a stirrer and thermometer .
The cells set up in this way , although made by different students-with different solutions at different dates , were found to agree more closely among themselves than the portable forms , owing probably to the more constant and certain conditions to which they were exposed .
They still , however , exhibited irregular differences , even when exposed to precisely similar variations of temperature , and it was felt that they could not be used with confidence for any work in which an accuracy of one part in 10,000 was desired .
Some determinations were also made of the temperature coefficients of these cells when exposed to a variation of temperature at the rate of about 10 ' in two hours .
The results were very fairly consistent for each cell , but gave very different values for the mean coefficient between 10 ' and 20 ' C. The lowest value obtained was 0-00045 , the highest 0-00069 .
The value commonly given for these cells is 0-00078 .
They were all saturated cells containing an abundance of crystals , which remained visible on the top of the paste and throughout the mass at the highest temperature .
Electromotive Force of different Forms of the Clark 119 In examining the results it was noticed that the value of the coefficient did not depend on the quantity of crystals in the cell , but on the distance of the end of the zinc rod from the crystals .
The difference of the results was , therefore , evidently due to slowness of diffusion of the state of saturation through the body of the solution .
So long ago as 1886 a set of five cells was set up by one of the present writers at the Cavendish Laboratory , Cambridge , with zinc rods of different lengths , with the object of testing the effect of diffusion on the temperature-coefficient .
These five cells were of the original Kayleigh pattern , with platinum wires sealed through the bottom .
They could not be immersed in a water-bath , and were found to be of a somewhat unsuitable form for the experiment , which was then discontinued for more important work .
Some more recent tests of this set of cells will be found in the paper on the Clark cell , by Glazebrook and Skinner.* The cells are numbered , 6 , 7 , 8 , 9 , and 10 in the paper , and are among the oldest in the possession of the Cavendish Laboratory .
It will be seen from the tests that they exhibit small discrepancies such as might very probably arise from differences in the time required for diffusion in the different cells .
From a study of the above-mentioned paper , it appears likely , in our opinion , that the differences observed in the case of the other cells of the same type may have been affected by a similar cause .
It would appear , in fact , inevitable that cells of the simple Board of Trade pattern should exhibit some effects of diffusion-lag , especially if subjected to considerable or rapid changes of temperature .
This form of cell , however , is so convenient to use , and so easy to make , that it seemed to us desirable to make a more careful examination of the case , with the object* if possible , of constructing cells of this form with a definite temperature-coefficient and a negligible diffusion-lag .
In October , 1894 , the class was joined by Mr. H. T. Barnes , who , as assistant to Dr. Harrington , Professor of Chemistry , had obtained considerable experience in chemical manipulation .
From this date onwards the work of making and testing various forms of Clark cell has been performed almost entirely by Mr. Barnes , but the observations and calculations throughout have been checked and verified by Professor Callendar , who has devoted special attention to the thermometry .
S 3 .
Constant-temperature Baths .
The first step in the investigation consisted in making a pair of suitable water-baths , controlled by delicate thermostats , which could * \#171 ; Phil. Trans. , ' A , vol. 183 ( 1892 ) , p. 586 .
Prof. H. L. Callendar and Mr. H. T. Barnes .
be set in such a manner as to keep the temperature steady for any desired length of time at any point between 5 ' and 30 ' C. The method of regulation adopted for these baths was similar to that described by Griffiths , * but much less elaborate .
The baths were made of copper , and were encased in felt and wood .
They were heated by a stream of tap-water passing through a copper tube over a regulated gas flame .
The regulators were made to cut off the gas so sharply that a difference of temperature of one-tenth of a degree sufficed to change the gas supply from full flame to no flame .
With these very sensitive regulators , some trouble was experienced at first , partly owing to the excessive variations of the Montreal gas pressure , and partly owing to the sudden changes of the climate. .
In the end , however , these difficulties were so successfully overcome that , on the longest continuous run , extending over nearly a fort-night , the temperature of the bath did not vary by so much as 002 ' C. throughout the whole period .
The temperature of the tap-water averaged about 8 ' C. in mid-winter , and seldom rose above 13 ' 0 .
in summer .
This generally sufficed to keep the baths steadily at 15 ' C. even if the temperature of the room was as high as 25 ' C. In order to obtain steady temperatures at points below 15 ' 0 .
, the stream of tap-water was led through a copper spiral immersed in melting snow or ice before passing over the gas flame .
In this manner the baths could be set to regulate steadily at temperatures as low as 5 ' C. The two baths were generally set to regulate at different temperatures , so that by transferring a cell from one bath to the other the effect of a sudden and definite change of temperature could be observed .
The time required by cells of different forms to reach their steady final values could thus be determined , and the effects of diffusion-lag could be readily distinguished from the immediate change of E.M.F. consequent upon a change of temperature .
S 4 .
Electrical Apparatus .
By the use of these accurately regulated water-baths , the temperature of the cells became so much a matter of certainty that we found it desirable to make the comparisons to the hundredth of a millivolt , corresponding to the hundredth of a degree Centigrade of temperature .
From the results of our experiments , we have reason to conclude that the Clark cell , under suitable conditions , permits the attainment of this order of accuracy , and is far superior to the silver voltameter for accurate determinations .
The comparisons of the cells were made by the usual Poggendorff method , with a 6000-ohm galvanometer .
The potentiometer used * * Phil. Trans. , ' A , vol. 184 ( 1893 ) , p. 374 .
Electromotive Force of different Forms of the Clark Cell .
121 for the earlier comparisons was a long wire , having a resistance of 86 ohms , wound on a cylinder in one hundred turns .
Bach turn was divided into one hundred parts , and readings were taken to one tenth of a division .
Each division corresponded approximately to one five-thousandth part of a volt , a storage-cell being used to supply the steady current through the wire .
This potentiometer had been accurately calibrated throughout its length at two different dates .
The results agreed so closely that it could be used with confidence for measuring relatively large differences of potential with an accuracy of at least one-half division of the wire , equivalent to 0-0001 of a volt .
The errors of the uncorrected wire amounted to over ten divisions in many places .
Bor the later experiments a simpler , and in many respects more convenient , form of potentiometer was used .
Two resistance boxes , containing resistances adjustable up to 2000 ohms each , were connected by .a platinum-silver bridge-wire having a resistance of 18 ohms .
The wire was 2 metres long , in four lengths of 50 cm .
each , with a millimetre scale , and was adjusted to read direct in volts , at the rate of 1 mm. to one-hundredth of a millivolt , in the following manner:\#151 ; A resistance of 18/ 20 X 1420 ohms was taken out of the first box , and the resistance in the second box was adjusted to make the standard cell at 15 ' C. read near the point 140 cm .
of the wire , i.e. , 14 millivolts above T420 volts .
The bridge-wire could thus be used directly for measuring small differences of E.M.F. not exceeding 20 millivolts , with an accuracy of at least one part in a thousand on a difference of this order .
Larger differences could be readily dealt with by transferring resistance from one box to the other in such a way as to keep the sum constant , each ohm transferred being reckoned at 20/ 18 of a millivolt .
It will be understood that the resistance of the bridge-wire was carefully measured in terms of the boxes , that their temperature-coefficients were very nearly the same , and that the wire was tested for uniformity , to insure the above order of accuracy in the determination of differences of E.M.B. of this magnitude .
S 5 .
Thermometry .
In working to the hundredth of a millivolt , it was necessary to \#177 ; cnow the temperature of the baths to the hundredth of a degree C. Two thermometers were generally used , one in each water-bath .
They were both carefully compared with a platinum thermometer , and their indications were in all cases reduced to the absolute scale .
One of the thermometers was by Geissler , divided to tenths of a degree .
This thermometer had evidently been graduated to read temperature on the absolute scale direct .
Its errors , after correcting -Piof .
IX .
Ij .
Callendar and JVlr .
H , T , Barnes .
for rise of zero , were found to be very small and irregular , seldom exceeding 001 ' C. The second thermometer was by Hicks , divided to twentieths of a degree .
Its corrections were found to be very nearly the same as those of the Kew mercurial standard .
Over the range 0 ' to 30 ' C. the changes of zero of these thermometers would never exceed 0 01 ' C. , and were , therefore , disregarded , ihe correction for the length of stem exposed , never exceeding two or three hundredths of a degree , could be applied with sufficient accuracy when required .
The comparisons were made with a platinum thermometer constructed ox special wire , which has been repeatedly tested by Professor Callendar , and also by Mr. Griffiths , and by Messrs. Heycock and Neville .
The wire is the same as that used in the thermometers made for the Kew Observatory , and its " delta-coefficient " has been taken as P50 .
The resistance box used was of special design , reading to 0'0001 ' C. It was exhibited by Professor Callendar at the May Conversazione of the Royal Society in 1893 .
The box used at Kew for platinum thermometry has recently been constructed on the same model , and has been described in ' Nature , ' November 14 , 1895 .
The Kew box differs chiefly in the use of plugs for mercury contacts , and in the absence of the temperature compensating coils .
S 6 .
Comparisons of Board of Trade Cells .
With this apparatus many more accurate and careful comparisons of Board of Trade cells were made .
Several new cells , prepared by H. T. B. and by other students , were compared with those a year old .
The newer cells were generally found to have a slightly higher E.M.F. than the old , and in general differences of the same order as before were observed , if the cells were subjected to different treatment .
It was noticeable , however , that B.O.T. cells , prepared about the same time in a similar manner , if kept exposed to similar stable conditions , would generally attain the same E.M.P. , within one or two tenths of a millivolt , after a day or so in the constant temperature bath at 15 ' C. The importance of keeping cells of this type at a constant temperature has been shown by Griffiths , * who has obtained very consistent results with B.O.T. cells treated in this manner .
Kahle , on the other hand , f finds differences , amounting to 4 or 5 millivolts in some cases , between the nine B.O.T. cells which he tested under constant and similar temperature conditions .
Such differences are quite beyond the range of our experience , and we do not * * Phil. Trans./ A , vol. 184 ( 1893 ) , p. 388 .
t ' Wied .
Ann./ vol. 51 ( 1894 ) , p. 194 .
Electromotive Force of different Forms of the Clark 123 think that they fairly represent the performance of B.O.T. cells under the conditions that he describes .
S 7 .
Temperature Coefficients of B.O.T. Cells .
Fresh determinations of the temperature coefficients of the old cells , in addition to those of the new cells , were made under different conditions and with greater accuracy by means of the constant temperature baths .
For four of the cells in which the end of the zinc rod was at a small distance from the crystals , the mean coefficients obtained on raising the temperature from 15 ' to 25 ' C. varied from 000046 to 0*00051 , and were in practical agreement with previous tests of the same cells under similar conditions .
Precisely similar values were obtained after keeping the cells at 25 ' C. for the night , and then lowering the temperature to 15 ' C. In another case , after keeping two of the cells at 25 ' C. for the night , the changes of their E.M.F. from their values at 15 ' C. were found to be 9*0 and 8*6 millivolts , giving co-efficients 0*00063 and 0*00060 respectively .
The greater change is evidently due to the time allowed for diffusion .
On cooling the cells down from 15 ' to 0 ' C. , allowing them to remain for an hour at the latter temperature , the mean coefficients obtained were invariably much larger .
The reason is evidently that the zinc becomes partially imbedded in the crystals at the lower temperature , and is necessarily in contact with a normal saturated solution throughout a considerable portion of its surface , In one -case a coefficient as high as 0*00075 was obtained ; in another a coefficient as low as 0*00059 .
In the latter case the rod was very long , and a considerable length was probably exposed to a supersaturated solution .
That this state of supersaturation is likely to occur , and to persist for a considerable time , is also illustrated by another experiment with a cell containing very few crystals .
After keeping the cell in question at 25 ' C. for a day , it was observed that all the crystals had disappeared , whereas the other cells still showed considerable quantities .
On cooling the cell down to 15 ' C. , the E.M.F. rose with a coefficient 0*00040 , and then remained steady for some time , no crystals reappearing .
After a time a sudden rise in the E.M.F. was observed , and the crystals were seen to have reappeared on the surface of the paste .
On transferring the cells back from the melting ice to the constant temperature bath at 15 ' C. , the E.M.F. of the B.O.T. cells was almost invariably found to be from 2 to 4 millivolts higher than before -cooling .
The difference was greatest in those cells which contained the greatest quantity of solution , and persisted for several days if Prof. H. L. Callendar and Mr. H. T. Barnes .
the cell were not disturbed .
In one case , after cooling to 0 ' C. , a cell remained nearly 5 millivolts too high when kept for one hour at 15 ' C. After twenty hours at 15 ' C. , the difference still remained 3 millivolts .
In the course of the next two hours it was shaken twice , and returned to within one-tenth of a millivolt of its previous value .
On further shaking the cell it was noticed that , if the mercurous sulphate were disturbed so as to come in contact with the zinc , the E.M.I. temporarily fell some 2 or 3 millivolts , but recovered very quickly on the mercurous powder subsiding .
This observation illustrates the necessity , now well understood , of keeping the zinc from direct contact with the mercurous paste .
S 8 .
Illustration of Diffusion-lag .
To illustrate the extreme slowness of the diffusion process , and to show that all saturated cells have really the same coefficient , if sufficient time be allowed for diffusion , the following experiment was tried .
Two exactly similar Board of Trade cells of normal E.M.F. were taken from the constant-temperature bath at 14 ' C. , and placed in the other bath at 25 ' 0 .
Two or three observations were taken each day of their subsequent changes of E.M.E. After the lapse of two days , one of the cells was occasionally shaken , and rapidly gained the correct value of the E.M.F. of a saturated cell at 25 ' C. The other cell was left undisturbed , the temperature being maintained constant to one-fiftieth of a degree Centigrade .
The E.M.F. of the latter cell fell slowly and almost uniformly as the diffusion proceeded , but it was not until after the lapse of nearly a fortnight that it reached the correct value .
The annexed curves ( Fig. 1 ) illustrate the rate of diffusion in these cells .
The absciss\#174 ; represent time in days ; the ordinates , fall of E.M.F. in millivolts from 15 ' C. For the sake of comparison , a cell of a different type , which we designate " B.O.T. crystal , " was submitted to the same treatment at the same time .
The straight line BC relates to this " crystal " cell .
It will be seen that it shows no appreciable diffusion-lag .
In fact its E.M.F. had arrived in twenty minutes within a tenth of a millivolt of its final value .
Curve No. ( 1 ) relates to the B.O.T. cell which was shaken .
The points at which the shaking took place are marked by the sudden falls of E.M.F. Curve No. ( 2 ) relates to the cell which was left undisturbed .
The rate of diffusion would probably have been much slower and more uniform , but for the slight vibration due to the running of the stirrer in the water-bath .
The changes in the rate of diffusion shown at the points 2'5 days and 8'5 days were probably due to excessive stirring about those dates .
Electromotive Force of different Forms of the Clark Cell .
Fig. 1.\#151 ; Curves showing Diffusion-Lag of two B.O.T. Clark Cells .
We have also tested several of the Muirhead portable cells in a special water-jacketed air bath , the temperature of which was regulated by a thermostat .
We find that for comparatively rapid changes of temperature , such as 10 ' 0 .
in two hours , they have a temperature-coefficient of 0*00050 , on the average , between 10 ' and 30 ' C. They have also a slow diffusion-lag similar to other B.O.T. cells .
It must be remembered , however , that these cells , owing to their form , are not intended for the most accurate laboratory tests , and that they are quite sufficiently constant for the purposes for which they are generally used .
It is sufficiently evident from the examples above given that a Clark cell of the B.O.T. form containing clear solution cannot be said to have a definite temperature coefficient .
The change of E.M.F. is seen to depend on the previous history of the cell , on the rate of change of temperature , and on the quantity of solution and relative size and position of the zinc rod .
The temperature-coefficient at 15 ' C. may have any value , from 0*00040 to the value 0*00078 , which is generally taken .
If the latter value of the coefficient is regularly used , errors which are relatively considerable in accurate work may readily be made , even if the rate of change of temperature is only 4 ' or 5 ' a day .
There is no doubt that , provided suffi12b Prof. H. L. Callendar and Mr. H. T. Barnes .
cient time is allowed , and sufficient crystals are present , the higher value is the more correct one to use ; but we are inclined to think that many observers are not aware how extremely slow the process of diffusion really is , and how considerable a time is required for the attainment of the limiting value .
S 9 .
Board of Trade " Crystal " Cells .
The defects of the ordinary B.O.T. form of Clark cell in this respect of diffusion-lag have long been recognised , and various methods have been proposed to remedy the disadvantages resulting from it .
Lord Rayleigh himself preferred the H-form of cell , in which the zinc is buried under a layer of crystals .
This form of cell has been adopted by the Berlin Reichsanstalt , and has been shown by Kahle to be practically free from the defects of the other pattern .
We have made several cells of the H-form , but we are strongly of opinion that , besides being more difficult to make , they are not so convenient to work with as the B.O.T. test-tube pattern .
On considering the matter in the light of the preceding observations with regard to the great differences in the temperature-coefficient produced by different degrees of immersion of the zinc rod in the crystals , it occurred to us that the diffusion-lag of the B.O.T. pattern might be entirely removed by a very simple modification of procedure ; so simple , in fact , that it would seem scarcely to deserve mention , were it not that of the many hundreds of B.O.T. cells which we have seen and examined , not one has been constructed in the manner to be described .
The modification we have adopted in these cells , which we term Board of Trade " crystal " cells , consists simply in filling the cell above the mercurous paste with moist crystals , instead of with saturated solution .
Under these circumstances no part of the solution can remain either supersaturated or unsaturated for any appreciable time .
We have subjected these and other cells to the severest tests , and the most sudden variations of temperature , such as 0 ' C. to 25 ' or 30 ' C. , and we find that the B.O.T. cells , when filled in this manner with crystals , have no appreciable diffusion-lag , and are not surpassed in quickness by any other form .
S 10 .
Preparation of Crystal Cells .
The procedure which we adopt in making these crystal cells differs only in one or two small details from that prescribed in the Board of Trade memorandum on the Clark cell .
A stock solution with mercurous sulphate paste is prepared Electromotive Force of different Forms of the Clark Cell .
127 exactly as therein described .
When cool , the supernatant liquid may be decanted off to be used for the preparation of crystals .
A supply of suitable crystals is easily obtained at any time by taking this solution , or any other solution of zinc sulphate -which has been neutralised and treated with mercurous sulphate at 30 ' C. , and cooling it down to 0 ' C. The liquid is then decanted off , and the crystals drained on a piece of filter-paper .
This method leaves them sufficiently moist for the purpose .
The crystals are filled into the cell through a glass tube or funnel to a depth of about 2 cm .
, and a hollow is made in the surface with a glass rod to facilitate the introduction of the zinc .
When the cell has settled it should be free from air-bubbles , and should show the merest film of liquid above the surface of the crystals .
Inasmuch as zinc sulphate does not tend to form any hydrate higher than the hepta-hydrate between the limits 0 ' and 30 ' C. , the crystals will remain equally moist , or very nearly so , , between these limits .
After soldering the platinum wire on to the zinc rod , we prefer to seal the end of the zinc rod with marine glue into a glass tube which nearly fits it .
The object of this is to make the best possible seal to protect the solder joint , which may otherwise be injured by the creeping of the solution .
The upper part of the glass tube is then sealed on to the platinum wire to form a mercury cup .
The glass tube also forms a convenient handle to use when inserting the cork and the zinc rod into the cell .
We prefer to amalgamate the zinc rod , as this proceeding appears to protect the zinc from local action , and to give more uniform results .
We may here remark that in the B.O.T. cells containing solution the zinc rod , even if amalgamated , rapidly becomes corroded near the top by local action .
In the cells filled with crystals , on the other hand , the zinc remains perfectly bright and clean .
S11 .
Decomposition of Mercurous Sulphate .
After neutralising the zinc sulphate solution with zinc oxide , we prefer to filter in a jacketed funnel at 40 ' C. , for the sake of obtaining a stronger solution .
Mercurous sulphate is then added to remove any traces of zinc oxide or other impurities which have any action upon it .
We have observed that if , after the addition of the mercurous sulphate , the solution be heated to between 35 ' and 40 ' C. , a slight change may be noticed in the appearance of the mercurous sulphate .
The filtrate , when cooled to 15 ' C. , may remain clear , but , if further cooled to 0 ' C. , a yellow turbidity makes its appearance , , showing that the mercurous sulphate has , in all probability , been partly decomposed by exposure to the higher temperature .
If , on the other hand , the zinc solution has not been heated above 30 ' C. 128 Prof. H. L. Callendar and Mr. H. T , Barnes .
with the mercurous sulphate , the filtrate will remain clear when cooled to 0 ' C. It appears probable that if a Clark cell is heated above 30 ' C. , or is made from paste which has been so heated , its E.M.E. may be affected by a similar cause .
The temperature of 30 ' C. appears , however , to be a perfectly safe limit .
It is not unlikely that the decomposition is determined by the presence of zinc oxide , as we have rarely observed changes of more than one or two tenths of a millivolt even after heating a cell to 50 ' C. S 12 .
Tests of B.O.T. Crystal Cells .
Several of these crystal cells were made by H. T. B. , and later by the advanced electrical students in the ordinary course of their work .
The cells so set up at different times by different students were rarely found to differ under any conditions by so much as the tenth of a millivolt from the mean at any given temperature .
If , as occasionally happened , a new cell , within an hour or so of sealing up , was found to have an E.M.P. as much as two tenths of a millivolt too high , it was short-circuited for half an hour or so with a piece of copper wire .
This procedure invariably had the effect of reducing the E.M.E. to its normal value .
It might naturally be supposed that with so small a quantity of solution , these cells would be seriously affected by short-circuiting .
We have found , on the contrary , that they are much less affected than the ordinary B.O.T. form or than unsaturated cells .
The crystal cells on short circuit were found to give a current of about 5 or 6 milliamperes , falling gradually to 2 or 1 in the course of an hour .
On removing the short circuit the cells ilistantly recovered to within a millivolt of their normal value , so quickly , in fact , that it was found impossible by the balance method to obtain any intermediate readings showing the rate of recovery .
In less than five minutes the value had generally recovered to within a tenth of a millivolt of the normal .
The ordinary B.O.T. cells of the same size , containing clear solution , were found to give a similar current on short circuit , but the recovery was never so rapid or perfect .
S 13 .
Temperature Coefficient of Crystal Cells .
Having satisfied ourselves by various preliminary trials that the crystal cells were practically free from diffusion-lag , and finding that the temperature coefficient between 15 ' and 0 ' C. appeared to be somewhat higher than that given by Kahle and other authorities , we determined to make a systematic series of observations under definite and uniform conditions .
Electromotive Force of different Forms of the Clark Cell .
129 For this test four crystal cells of different dates were selected , differing as widely as possible ( one-tenth of a millivolt either way ) from the mean at 15 ' C. Six other cells were set apart as standards of comparison , and were kept at a constant temperature of 15 ' C. night and day throughout the series of observations .
The cells wei'e all of the same form and dimensions as the B.O.T. cells above described , and were all sealed with marine glue .
At starting the mean of the test cells at 15 ' C. agreed to one hundredth of a millivolt with the mean of the standard cells .
The four test cells were then immersed in melting snow , and comparisons of their E.M.F. with the standards were made at intervals of six , eighteen , and twenty-four hours after immersion .
For the next twenty-four hours they were kept at a temperature of 5 ' C. , next day at 10 ' C. , then for three days at 15 ' C. , then for a day each at 20 ' , 25 ' , and 30 ' C. In this manner the comparisons at each point were made as nearly as possible under similar conditions .
The small differences which the cells possessed at starting were maintained , within two or three hundredths of a millivolt , throughout the series of observations .
They were possibly due to inequalities of age , or to slight differences in the preparation of the solutions and crystals , or to the fact that the platinum wires in the mercury cups were not amalgamated .
No systematic difference in their temperature-coefficients could be detected .
At each point the mean of the observations taken at the end of the first six hours showed a slight lag , as compared with the observations taken at eighteen and twenty-four hours , amounting on the average to nearly two hundredths of a millivolt .
It is necessary to remark , however , that a difference of one hundredth of a degree of temperature means rather more than one hundredth of a millivolt ; and that this apparent lag , corresponding to less than 0'02 ' 0 .
, may have been due to some constant error of observation of temperature under different conditions in the morning and evening .
In any case it is evident that the diffusion-lag , if any , is so small that it would be quite useless to consider it in any case , unless the greatest precautions were taken to secure a constant and uniform temperature , and to measure the temperature to at least O'Ol C. After the three days at 0 ' , 5 ' , and 10 ' C. , the mean of the four cells returned in less than twenty-four hours to within one hundredth of a millivolt of the mean of the standard cells .
After three days at 15 ' C. the mean values were identical .
At the conclusion of the observations a.t 30 ' C. , the cells were replaced in the bath at 15 ' C. In fifteen hours the mean value had returned to within two hundredths of a millivolt of the standards .
Some two months after the above series of experiments , a second set of observations at 30 ' C. was taken with three of the same cells , by 130 Prof. H. L. Callendar and Mr. H. T. Barnes .
way of verification .
The cells at 15 ' C. were still found to preserve their relative differences to within one or two hundredths of a millivolt , and the mean still agreed with that of the standards .
The difference at 30 ' G. was observed directly in terms of the resistance boxes , as well as in terms of the bridge-wire , at each observation .
The fall of E.M.F. found was less by four hundredths of a millivolt in I9'4 millivolts , than on the preceding occasion .
After twenty-seven hours at 30 ' C. the cells , when replaced in the 15 ' C. bath , returned in twenty hours to within two hundredths of a millivolt of their previous values .
S 14 .
Quickness of Recovery .
It must not be hastily assumed from the foregoing observations that the cells in each case took nearly a day to recover their original values at 15 ' C. The observations were taken in this particular series after the lapse of several hours in order to make sure that the cells had not undergone any permanent change as the result of prolonged exposure to 0 ' and 30 ' C. We have made several special tests with regard to this point .
We find that B.O.T. crystal cells of this size , set up in test-tubes nearly 2 cm .
in diameter , when suddenly transferred from melting snow , or from the other bath at 30 ' , or even 40 ' C. , back to the constant temperature bath at 15 ' C. , return to within a tenth of a millivolt of their previous values in less than ten minutes .
For cells of a smaller size , of which we have made several , the recovery is still more rapid .
It appeal 's to be chiefly a question of the time required for the change of temperature .
After exposure to a temperature above 15 ' C. , the cells frequently return , at 15 ' C. , in less than half an hour to within one or two hundredths of a millivolt of their previous values .
After exposure to a temperature below 15 ' C. , the recovery is a little slower .
This might naturally be expected , as crystallisation is generally more rapid than solution .
We have also taken a series of observations at 40 ' C. , though this temperature lies outside the limits of practical utility .
On suddenly raising the temperature to 406 ' C. the value observed after ten minutes was one millivolt higher than the final value .
The next observation was taken after three hours , by wdiich time it was found , on subsequent redaction , that the E.M.F. had become constant .
The E.M.F. at 40'60 ' C. was found to be 35'81 millivolts lower than at 15 ' C. The recovery on returning to the 15 ' bath after a day at 40'6 0 .
was quite unexpectedly rapid .
In ten minutes the value was found to be within a tenth , in two and a half hours within a hundredth of a millivolt .
Later observations showed no further change .
Electromotive Force of different Forms Clark Cell .
131 S 15 .
Results of Observations .
Table I.\#151 ; Temperature-variation of B.O.T. Crystal Cells .
Temperature C. corrected to nitrogen-scale .
Difference in millivolts from value at 15 ' C. Difference from lineality .
Difference from formula ( p ) .
Observed ( O ) .
Formula ( L ) .
Observed ( O)-(L ) .
Calculated formula ( p ) .
o-oo ' +16 -62 +18 -00 -1-38 -1-40 + 0-02 5-17 + 11-15 + 11-80 -0-65 -0-60 -0-05 9-89 + 5-92 + 6-13 -0-21 -016 -0-05 20-43 - 6 70 - 6-52 -0-18 -0-18 -o-oo 24-75 -12 -25 -11 -7o -0-55 -0-59 + 0-04 29-90 -19-42 -17-88 -1 '54 -1-38 -0-16 29-86 -19-32 -17-83 -1-49 -1-37 \#151 ; 0 .
] 2 40-60 -35-81 -30-72 -5-09 -4-07 -1 -02 Fio .
2 .
Sca/ e of Tempera .
Cure .
um L/ nkar The results of these experiments on the temperature-variation of the E.M.F. of B.O.T. crystal cells are given in the above table , and are plotted in the accompanying curve ( fig. 2 ) .
The curve is drawn to show not the whole temperature-variation of the E.M.F. , but the defect of the change from lineality .
The observed E.M.F. in volts at a temperature t ' C. is less at all points than that calculated by the simple linear formula , E , = E16-0-001,200 ( \#163 ; \#151 ; 15 ) ... ... ... ... ... . .
( L ) .
The difference from this linear formula is approximately repi-e-sented by the addition of a parabolic term , \#151 ; 0-0000062 ( \#163 ; \#151 ; 15)3 ... ... ... ... ... ... .
( P ) .
VOL. LX1I .
L Prol .
H. L. Callendar and Mr. H. T. Barnes .
The full curve in fig. 2 represents this difference-term on a scale of 2S- cm .
to 1 millivolt .
The crosses represent the results of actual observation at the different points .
The mean difference of the observations from the curve , if we except those at 30 ' C. , is only three-hundredths of a millivolt .
The observations at 30 ' C. differ by more than one-tenth of a millivolt , and those at 40 ' C. by a whole millivolt from the simple parabolic curve .
These differences cannot , we think , be explained as being due to errors of observation .
This is proved by the accuracy with which the cells returned to their original values at 15 ' C. , and also by the agreement of the twenty or thirty different readings at each point .
Moreover , the observations have been repeatedly verified by others , not shown on the curve , with cells of different types , to within one or two-hundredths of a millivolt .
We conclude that no simple parabolic formula can be made to fit the observations throughout the range 0 ' to 40 ' 0 .
Over the range 0 ' to 28 ' 0 .
, however , the differences , even if real , are not of great importance , and we may take the formula , E\lt ; = E15-0-001200 { t-15)-0-0000062 ( \#163 ; -15)2 , ... ... ... ... ... .
( L ) + ( P ) as representing the temperature-variation of the E.M.F. of these cells within about one-twentieth of a millivolt between these limits .
The symmetry of the curve shows that we may take the very convenient round number T200 millivolt , for the change of E.M.F. per 1 ' C. at 15 ' C. Taking formula ( P ) , we find for temperature-coefficient at t ' C. , d/ dt ( E , / E1# ) = -0-000837-0-0000087 0-15 ) , and for the mean temperature-coefficient between t ' and 15 ' C. , ( E , / E15\#151 ; 1)1 ( t\#151 ; 15 ) = \#151 ; 0-000837\#151 ; 0-0000043 ( t - 15 ) .
It is generally , however , more convenient to use the formula ( L + P ) .
Between the limits of 12 ' and 18 ' C. we may use the simple linear formula ( L ) , without risk of making an error greater than one-twentieth of a millivolt .
If , however , we were to use the temperature-coefficient 0"00076 ( which is very commonly taken ) over the same range , the error might amount to nearly three-quarters of a millivolt .
S 16 .
Results of Previous Observers .
The formula given above for the temperature-coefficient differs from that of previous observers chiefly in the direction of making the change of E.M.F. more nearly uniform .
Lord Rayleigh* tested two cells under similar conditions of slow temperature change .
For one cell he found the temperature-coefficient at t given by the formula 0-00083 + 0-000018(^ \#151 ; 15 ) , which * 'Phil .
Trans. , ' vol. 176 ( 18S5 ) , p. 794 .
Electromotive Force of different Forms of the Clark Cell .
133 agrees with our value , except that the rate of change of the coefficient is twice as great .
For the other cell he found a coefficient 12 per cent , smaller with a similar rate of change .
His cells , in which the zinc was pushed down into the paste , would be certainly less liable to diffusion-lag than the ordinary B.O.T. pattern , and might possibly possess a different coefficient ; but we think that the difference is most probably to be explained by diffusion-lag , and in that case the higher value would be the more correct .
Glazebrook and Skinner find the value 000076 for the mean \#166 ; coefficient between 0 ' and 15 ' C. , under conditions in which diffusion-lag would be approximately eliminated .
Our value between these limits is 0-000773 .
Our value for the coefficient at 15 ' C. , namely , 0-000837 , is also in fairly close agreement with the value 0*00082 for an H-cell at 15 ' C. , found by Fleming .
The lower values obtained by many observers for ordinary B.O.T. cells , are doubtless vitiated by the effects of diffusion-lag , and are , in this respect , in agreement with our own results for such cells under similar conditions as given in a previous section .
The observations of Kahle* are probably the most systematic .
He finds for the E.M.F. in volts at t ' C. the formula E* = E15\#151 ; 0 00116 ( * \#151 ; 15)\#151 ; 0-00001 0-15)2 ... ... ... ( K ) .
The difference between this formula of Kahle and the linear formula ( L ) is show-n by the dotted curve in fig. 2 .
It will be seen that the agreement with our observations is very close between 25 ' and 30 ' C. , but that the value of the coefficient at 15 ' 0 .
is somewhat smaller .
Below 10 ' C. the divergence is very marked .
The formula of Kahle gives a change of only 15*15 millivolts between 0 ' and 15 ' C. , corresponding to a mean coefficient of 0*000704 , values which are evidently much smaller than those given above .
We do not think , however , that it is necessary to assume that there is any real difference of behaviour between our cells and those tested by Kahle .
The discrepancy is more probably to be explained by the fact that the observations on which the formula of Kahle is founded were taken between the limits 12 ' and 28 ' C. , under conditions less favourable to the cells .
Between these limits , so far as we are able to judge , we are in agreement with Kahle , within the limits of accuracy of his observations .
Kahle does not give any detailed observations or any statement of the probable error of his results , but it is possible to form a general idea of the limits of accuracy from the account which he gives of his method .
The cells were kept immersed in paraffin baths , regulated by means of " Rohrbeck " thermostats .
The temperature seldom varied more than one degree from day to * ' Wied .
Ann. , ' vol. 51 ( 1894 ) , p. 197 .
L 2 Prof. H. L. Callendar and Mr. H. T. Barnes .
day in either bath , or by more than a tenth of a degree in the course of an observation .
From other observations given by Kahle* it would appear as though his cells were subject to a temperature or diffusion-lag of the order of half a millivolt , when the temperature was changing at the rate of 1 ' C. per hour .
From these and other considerations it is evident that Kahle did not aim at an order of accuracy higher than one or two-tenths of a millivolt , and that his formula could not be expected to give correct results beyond the limits of observation .
Thus , although his formula is practically correct between the limits of his observations , namely 12 ' and 28 ' C. , it is quite possible that it may be as much as 10 per cent , in error at 0 ' C. On the other hand , we regard it as quite impossible that our observations at this point should be in error by even a tenth part of the amount , namely 1^ millivolts , by which they differ from the formula of Kahle .
Again , although the observations of Glazebrook and Skinner between 15 ' and 0 ' C. may have been affected to a slight extent by diffusion-lag , it is plain that the effect of diffusion-lag , if any , must have been to reduce the extent of the change , and could not explain the fact that the change which they observed was so much larger than that given by Kahle 's formula , and so nearly in agreement with our own .
In this connection it is necessary to refer to an opinion which iv have often encountered in conversation and otherwise , and which is possibly still current , namely that the observations of Glazebrook and Skinner are in precise agreement with those of Kahle on this point .
For instance , Schusterf quotes correctly Kahle 's formula for the mean temperature-coefficient between and 15 ' C. , namely , a = 0-000814 + 0-000007 ( $-15 ) , and states that Glazebrook and Skinner 's coefficient ( = 0"00076 ) refers to a mean temperature of 7'5 ' C. , and is identical with the above at that temperature .
This is obviously true if we put 7"5 ' in the formula , but not if we put = 0 ' .
The mistake appears to have , arisen from the above formula having been inadvertently described by Kahle as being the temperature-coefficient at t , instead of the mean coefficient between t and 15 ' C. But although the words , " Fur eine beliebige Temperatur t , " used by Kahle , may , perhaps , be ambiguous , the complete formula ( K ) , from which the other is derived , leaves no possible doubt as to the true meaning .
J * Loc .
cit. , p. 199 .
f ' Phil. Trans. , ' A , vol. 186 ( 1895 ) , p. 458 . .
X Note added Sept , 20 , 1897 .
In a more recent number of \#163 ; Wied .
Ann. , ' Oct. , 1896 , Kahle states incidentally that he lias found by direct comparison the difference 16'6 millivolts between 0 ' and 15 ' , instead of 15"1 as given by his previous formula .
No details of observations are given .
Electromotive Force of different Forms of the Clark Cell .
135 S 17 .
Hermetically Sealed Cells .
We have always felt somewhat dissatisfied with the usual practice of sealing up standard cells with marine glue or paraffin wax .
Provided that the marine glue sealing is carefully and conscientiously performed , it may , doubtless , remain good for a considerable time , if the cell is not exposed to extreme variations of temperature .
We have not , however , ourselves succeeded in making the marine glue seal stand many repetitions of the 0 ' to 30 ' 0 .
treatment .
Cells which have been thus treated for a month or two have invariably shown some signs of creeping .
Except in extreme cases , this creeping does not appear to produce much effect on the E.M.F. of saturated cells , but in the case of unsaturated cells the effect is serious .
The set of twelve Carhart-Clark cells in our collection , though evidently prepared and sealed with the greatest skill and care , have all suffered from the creeping out of the solution in the lapse of two years .
One of the cells is now 5 millivolts below its normal value .
It is only fair to add that , owing to the extremes of the Montreal climate , they have been subjected to an annual temperature range of 5 ' to 27 ' C. , and that all the Muirhead cells in our collection are similarly affected .
We have succeeded in making several forms of hermetically sealed cells , and we are of opinion that such cells are much to be preferred as standards to those sealed in any other way .
The following are the principal varieties on which we have made experiments .
( 1 ) Cells of the H -Type with Zinc Amalgam.\#151 ; We prefer to make this cell in the form QQ of an inverted Y. Fine platinum wires are first sealed into the lower extremities of the inverted Y , the limbs are then filled with zinc amalgam and crystals of zinc sulphate , and with mercury and mercurous paste as usual .
When sufficient materials have been introduced , the middle leg is sealed off .
This inverted Y-form is much easier to make than the H-form , as it involves only one T-join .
In making these cells , we prefer to use lead glass tubing about 5 to 8 mm. in bore .
Hermetically sealed cells of this form were made many years ago by Wright* for the purpose of testing the effect of dissolved air on the E.M.F. W -Form.\#151 ; When intended for immersion in a water bath , the lower limbs of the inverted Y are continued upwards beyond the seal to a height of 4 or 5 inches forming a W. The upturned limbs are partly filled with mercury , and are used for making connections .
As the result of our experience with cells of this description , we are not inclined to recommend the use of cells containing zinc amalgam at temperatures above 25 ' C. or below 10 ' C. As Lord Rayleigh has observed , these cells show a very remarkable tendency * !
Phil. Mag. , ' vol. 16 ( 1883 ) , p. 28 .
lo6 Prof. II .
Ij .
Callendar and Mr. H. [ ' .
Barnes .
to crack in the platinum seal of the leg containing the zinc amalgam^ especially if exposed to low temperatures .
We do not think that they could be trusted to stand the Montreal climate .
Mery few of our cells constructed on this pattern have survived a month or two of the 0 ' to 30 ' treatment , if they contained more than a mere button of amalgam .
( 2 ) H -FormCells with Zinc Hod.\#151 ; We have generally found cells made with zinc rod to he more reliable .
The zinc rod is cast in a small glass tube of suitable size .
The platinum wire , previously enclosed in a capillary glass tube , is thrust into the fused zinc .
When cool , the glass mould is broken off , and the zinc rod cleaned and amalgamated , and introduced into one leg of the H. The other leg is partly filled with mercury to which connection is made by a platinum wire in a glass tube after the Board of Trade method .
The other materials are filled in as usual .
The two legs of the H are then fused up at the top , the upper portions serving as mercury cups .
We prefer to use tubes of very small dimensions , and to make the horizontal connecting limb as short as possible .
This cell is not very easy to make , owing to the double fusion on to the platinum wires after the materials have been filled in , when the cell cannot be inverted .
If , however , the tubes are made sufficiently small , it is a very convenient ' and sensitive form of cell .
( 3 ) Board of Trade Form with Crystals.\#151 ; The single tube form with the zinc rod cast in a similar way on to a glass capillary , is equally efficient if filled with crystals , and is much easier to seal .
There is only one tube to seal , and it is possible to get at it evenly from all sides .
There is generally no difficulty in keeping the two wires separate , provided that the capillaries through which they pass have been drawn sufficiently thick and strong .
Both the capillaries may be expanded into mercury cups at the top , or the outer tube itself may conveniently be used to form the mercury cujd for the zinc terminal .
Portable Form.\#151 ; This cell is still more easily made in a portable form , in which the mercury is replaced by amalgamated platinum .
The cell may then be made upside down , the difficult seal being made first , before the materials are filled in .
The process of making the cell is briefly as follows .
A platinum wire is sealed into a thick glass capillary with a small mercury cup at one end .
The wire is left projecting some 2 or 3 cm .
beyond the glass at each end .
One end is then hammered flat to serve as the positive element , and both ends are amalgamated by heating and plunging in mercury .
A platinum wire capillary without a mercury cup is cast into a small zinc rod , and the free end is amalgamated .
A glass tube of suitable size and length to form the cell is melted down in the middle till it Electromotive Force of different Forms the Clark Cell .
137 is of sufficient size to just admit the passage of the two capillaries from opposite sides .
The capillaries are then held in their proper relative positions while the seal is completed , and the free ends of the wires are coiled down in their respective mercury cups .
The lower-end of the tube is then drawn out slightly to facilitate the final sealing off .
The materials are filled in after the usual method , but in the reverse of the usual order .
Moist crystals of zinc sulphate are packed round the zinc till it is covered to a depth of about a centimetre .
After inserting about half the mercurous paste , the flattened and amalgamated platinum wire is coiled down into a suitable position , and more paste is added .
The end of the cell close to the point where it is to be finally sealed , is preferably filled with moist crystals instead of paste .
The object of this is to avoid leaving any of the paste close to the seal , where it might suffer decomposition from the heat in sealing off .
Any excess of solution is dried off with filter paper , and the narrow neck is then sealed off with a fine flame .
On the whole , we prefer this portable form of cell to any of the other forms we have tried .
The shape of the cell and its small size make it very convenient .
The seal is comparatively easy to make , and the narrow glass neck separating the cell from its connexions is also an advantage , as it diminishes the risk of any error of tempeia-ture arising from conduction along the tube .
We have not found that there is any advantage in using mercury as compared with amalgamated platinum .
The cell has a higher internal resistance , and gives a smaller current on short circuit , but the recovery appears to be equally rapid and complete .
3\ e cannot find any systematic difference in the electromotive force at any temperature .
If anything , the amalgamated platinum has the advantage over the mercury , as the purity of the mercury is then comparatively unimportant , and redistillation is unnecessary .
There does not appear to be any advantage gained , in our experience , by using a strip of platinum foil in place of the fine flattened wire .
S 18 .
Tests of Hermetically Sealed Cells .
We have similarly tested several saturated cells of the 11-foim , and various other patterns above described as hermetically sealed cells .
We find that they all show the same temperature change of E.M.F. as the B.O.T. crystal cells .
The agreement in nearly every case is within one or two-hundredths of a millivolt ev en at 0 and 30 ' C. , the limits of the range .
The largest divergence was found in the case of a portable cell of type ( 3 ) , with an amalgamated platinum wire in place of mercury .
In this cell , both at 0 ' and 30 ' C. , the difference from 15 ' C. 138 Prof. H. L. Callendar and Mr. H. T. Barnes .
exceeded by one-tenth of a millivolt that of all the other crystal cells we have tested .
Even in this case the cell was at least consistent with itself .
It was first tested at 15 ' , 30 ' , 15 ' , 0 ' , and 15 ' C. , allowing a day at each temperature .
It was a new cell , the first of this type which we tested , and its value at the time was nearly three-tenths of a millivolt higher than the standard .
The value found at 0 ' C. was 16-72 millivolts above , and at 30 ' C. 1972 millivolts below its value at 15 ' C. , instead of +16-62 , and -19*58 millivolts respectively .
The cell being of a very small and sensitive form , the test was repeated a few days later in the reverse order , allowing only half an hour at each temperature .
The values found in the second test were +16-71 and \#151 ; 19-70 millivolts respectively .
We thought at first that the discrepancy might be due to some inherent peculiarity of this type of cell .
We have since tested other cells of the same type and size , with results which agree to 0+2 millivolt with the B.O.T. crystal cells .
The agreement is not confined to points 0 ' and 30 ' C. For instance , the difference found at 24 ' C. was 0+5 of a millivolt less than that calculated by the formula ( P ) .
The B.O.T. crystal cells at this point show a difference of 0+4 millivolt from the curve ( P ) in the same direction .
We consider that the divergence of one-tenth of a millivolt in the case of this particular cell must be regarded as exceptional .
S 19 .
Importance of Constant Conditions of Temperature .
We have quoted the above test partly as an illustration of the kind of agreement between cells of different types which it is possible to attain with suitable cells under definitely known conditions of temperature .
In attaining this order of accuracy the chief difficulty lies in the certain determination of the temperature of the cells .
To attain a certainty of the order of 0+1 ' C. , the following conditions are necessary:\#151 ; 1 .
The cells should be of an elongated form , and should be deeply immersed in a bath of liquid , which is constantly and vigorously stirred .
2 .
The temperature must be kept constant to 0-01 ' C. , and the thermometer used must be read and corrected to the same order .
If the cells are not deeply immersed their temperature will be affected by external conditions .
If the liquid is not constantly stirred it will tend to be hotter at the upper surface , especially if the liquid be very expansible , like paraffin , and the bath be hotter or colder than its sui*roundings .
If the temperature is changing the cells will not be of an uniform temperature throughout , and will lag behind the thermometer , unless they happen to be of a smaller size .
Electromotive Force of different Forms of the Clark Cell , lot* The electrical conditions are not less important , but are more easily realised and maintained .
Our galvanometer is sensitive to much less than a millimetre of the bridge-wire ( one hundredth of a millivolt ) Great attention is paid to the perfection of the insulation , and to the avoidance of tbermo-electric effects , which may readily amount to more than ten microvolts .
We may here remark that in testing any new batch of cells it is quite impossible to tell , till the results are worked out , whether they are in agreement with others .
The many coincidences found cannot therefore be the result of bias on the part of the observer .
We think we may fairly claim for the Clark cell an order of consistency approaching one-hundredth of a millivolt in the temperature changes of its E.M.F. S 20 .
Clark Cells in which the Solution is of Constant Strength .
It is well known that Clark cells , not containing crystals , in which the solution does not change its strength with change of temperature , have the advantage of possessing a temperature-coefficient which is less than half that of the saturated cells .
They are also practically free from the effects of diffusion-lag , as the density of the solution is always nearly uniform .
The best known cell of this type is the Carhart-Clark cell , in which the zinc sulphate solution is chosen as being* saturated at 0 ' C. An error of 2 ' C. in the temperature at which the solution is saturated , will make an error of only one millivolt , approximately , in the E.M.F. of the cell .
We have prepared several cells of this type at different dates and in different forms , with separately prepared solutions .
In cells so prepared , of similar patterns , we have not as a rule found differences greater than two or three-tenths of a millivolt .
These differences were probably due as much to other cause as to difference of strength of solution .
W e have generally sealed the cells hermetically to avoid creeping of the solution , which has a tendency to lower the E.M.F. in the case of unsaturated cells .
It is eyident that these cells must , on the whole , be less accurately re producible than the saturated cells .
We have also found that they are more liable to undergo slight changes of E.M.F. as a result of short-circuiting , or of exposure to high or low temperatures .
They appear , in fact , to be less stable than the cells containing crystals .
We have also prepared experimental cells with solutions weaker than the cell saturated at 0 ' C. We have observed in these cells a similar instability , becoming more marked as the solution is weakened .
We are inclined to attribute this instability to a difference in solubility nr diffusivity of the mercurous sulphate in the weaker solutions .
140 Prof. H. L. Callendar and Mr. H. T. Barnes .
We also found it possible to obtain observations of the change of E.M.F. of cells saturated at 15 ' and 30 ' C. , under conditions in which the solution was considerably supersaturated .
The cell saturated at 15 ' C. was kept for several hours at 0 ' C. without showing any trace of crystallisation .
This cell agreed at 15 ' C. with the saturated cells , and gave very consistent readings throughout the range 0 ' to 30 ' C. The change of E.M.F. per 1 ' C. was found to be 0'567 of a millivolt between 0 ' and 15 ' C. , and 0560 of a millivolt between 15 ' and 30 ' .
As the whole change of E.M.F. between 0 ' and 30 ' C. was only 16-90 millivolts , and as the E.M.F. of the cell rose by one-tenth of a millivolt after keeping for six hours at 0 ' C. , the observations may be taken as showing that the temperature-coefficient of this cell , whether in the supersaturated or unsaturated condition , is practically constant over the range 0 ' to 30 ' C. This is in marked contrast with the case of the cells containing crystals .
The tests of the cells saturated at 0 ' C. were very fairly consistent , , but not quite so good as those of the cells saturated at 15 ' C. They showed a mean temperature change of E.M.F. per 1 ' C. of 0543 of a millivolt .
There was no decided evidence of any variation of the temperature-coefficient over the range 0 ' to 30 ' C. Carlxart gives the formula:\#151 ; E* = E15 [ 1 - 1-000387 ( *-15 ) +0-0000005 ( *-15)2 ] , which would make the temperature-coefficient diminish slightly as the temperature rises .
( Change of E.M.F. 056 my .
per 1 ' C. ) .
The tests on the weaker cells were much less consistent , owing to the instability of E.M.F. above referred to .
The results of the tests pointed to a mean change of 055 of a millivolt per 1 ' C. The change observed between 0 ' and 15 ' C. was sometimes greater and sometimes less than that between 15 ' and 30 ' C. , but there was no decided tendency either way .
After keeping for some time at 0 ' or 30 ' C. these-weak cells sometimes showed permanent changes amounting to as much as half a millivolt .
S 21 .
On the Density of Solutions of Zinc Sulphate .
A knowledge of the density of solutions of zinc sulphate is required in order to trace the relation between the changes of E.M.F. , which depend on change of strength and density of the solution .
This point has been investigated by two of Professor Carhart 's students , , the result of wrhose work has been published by Professor Carhart.* These observers find for a cell saturated at 15 ' C. an E.M.F. nearly five millivolts higher than that of a cell containing crystals , and a density which appears to be correspondingly low .
For this and * * Proc. Amer .
Elect. Eng. , ' 1892 , p. 615 .
Electromotive Force of different Forms of the Clark Cell .
141 othei ' reasons , we thought it would be desirable to repeat the determinations .
The observations of Lannoy , * though evidently undertaken with great care , did not extend to the case of solutions as dense as those occurring in Clark cells .
Other observers appear to have confined themselves chiefly to the case of very dilute solutions .
According to the views of Valson and Bender , which are quoted by Ostwald and other authorities , the density of a salt solution , such as zinc sulphate , may be additively deduced from the observed densities in the case of some standard solution ( .
, a solution of ammonium chloride ) , by means of two moduli representing the acid and the base respectively .
We have calculated the densities according to the values which they give for the moduli at 18 ' C. , but it appears that the results are only a rough approximation , and miss what seems to be the most characteristic feature of the change of density .
In determining the relation between density and strength of solution , the chief difficulties to be encountered are in the exact measurement of the strength .
If the composition of the solution is determined by weighing out known quantities of the hydrated salt into a litre flask , it is very possible that errors may arise from evaporation or efflorescence , or from the presence of other hydrates .
In order to avoid these possible errors , we adopted the much more laborious method of evaporating a known weight of solution to dryness at 100 ' C. , assuming that the residue was the monohydrate .
Two determinations were made in this manner for each solution tested , and in addition , two control experiments were made in which the strength of the solution was measured by estimating the sulphate by means of barium chloride .
The following table contains the results of these determinations for seven different solutions .
Table II.\#151 ; Density of Zinc Sulphate Solutions .
Z11SO4 per cent , of solution .
( p. ) Density of solution ( d ) .
ZnS04 gram per c.c. pd/ 100 .
Difference 0 *9982 4-to \#151 ; cl. Calculated by ( t\gt ; ) formula .
6 35 1 -0653 0 -0677 0 -0006 o-oooo 8*46 1 -0896 0 -0923 0 -0009 0 -oooo 13-49 1-1522 0-1557 0-0017 0 -0006 17-09 1 -2020 0 -2130 0 -0092 0 -0070 23 -75 1-2872 0 -3062 0-0172 0 -0174 27 -27 1 3418 0 -3667 0 -0231 0 -0242 33-21 1-4400 0 -4790 0 0372 0 -0307 * Ostwald , ' Zeit .
Phys. Cliem .
, ' Nov. , 1895 .
Prof. H. L. Callendar and Mr. H. T. Barnes .
The first column contains the percentage by weight , ^ , of ZnS04 in grams per 100 grams of solution , as deduced from the observations on each solution .
The second column contains the values of the density , d , at 20 ' C. , obtained by weighing a special form of pipette carefully filled with the solution .
The third column gives tv , the weight of ZnS04 in grams per cubic centimetre of solution , and is obtained by dividing the product of the numbers in the first two columns by 100 .
If we add to this weight the number 0"9982 , representing the weight of water in grams per c.c. at 20 ' C. , and subtract the observed density , d , of the solution , we obtain as the difference given in the fourth column , the weight of water displaced per c.c. by the zinc sulphate in solution .
The observations of Lannoy reduced on a similar plan are as follows :\#151 ; ( p ) ( rf ) at 15 ' C. pdjlOO .
Difference O'9992 +io\#151 ; d. Calculated .
2-25 1*0226 0-0230 -0-0004 0 -oooo 5*60 1-0596 0 -0594 -0-0010 o-oooo 11-21 1 -1238 0-1250 + 0-0004 o -oooo 16*85 1*1949 0-2013 + 0-0056 + 0-0057 These observations , taken in conjunction with our own , would appear to indicate a simple relation between the density and the composition , of a kind which so far as we are aware has not been previously observed .
Up to a density of about 1'150 , the solution of zinc sulphate appears to take place approximately without change of volume .
The added molecules of ZnS04 do not appear to displace any of the molecules of water , so that the density at 20 ' C. is very nearly 0*9982 + w. Beyond this point , it appears that each added molecule of ZnS04 displaces one molecule of water , so that the density of the solution is very approximately represented by the expression d = 0*9982 + w-18(w-0*150)/ 161 ... ... ... ( D ) .
The nature of this relation is perhaps more clearly shown by the curves given in fig. 3 .
In this figure , the values of to are taken as abscissae , and the corresponding values of the difference 0*9982 + w \#151 ; d , as ordinates .
The sharp break which occurs at the point to = 0*150 is very clearly shown both by the observations of Lannoy , which are represented by crosses , and by our own which are represented by the dots enclosed in circles .
Those of Lannoy unfortunately do not extend far enough to afford a satisfactory verification throughout the range , but we have no reason to distrust our Electromotive Force of different Forms of Clark 143 own observations at the higher points , as they were all carefully verified .
The dotted curve , which is practically a straight line , represents the formula of Bender and Valson , which smooths out the break .
It is not theoretically improbable that a simple relation of this type should be found to hold approximately in the case of salt solutions .
At the same time it is hardly to be expected that any such simple expression should represent accurately the changes of density at all temperatures .
The expansion of water is anomalous in the neighbourhood of the freezing point , and the coefficients of expansion of solutions differ considerably from that of water at low tem-peratures , and generally increase with increase of strength of solution-These variations in the coefficients of expansion may well introduce secondary effects of a corresponding order in the changes of density .
In comparing the observations of Lannoy with our own , which were taken at a slightly different temperature , it would appear not improbable that systematic differences of this kind may exist , but the point obviously requires much more careful investigation , as the differences shown are so small , and might readily be explained by errors of observation .
For instance , at the two lowest points the density according to Lannoy is greater than 09992 + w. Since he apparently determined the composition of the solution by weighing out quantities of the heptahydrate , the discrepancy might be explained by a slight degree of efflorescence of the sample used for these deter-minations .
In the table of densities given by Carhart , the composition of the solution in each case is stated in terms of the percentage of ZnS04 by weight in 100 parts of solution .
If we assume , in the absence of any definite statement , that the symbol ZnS04 stands in this case also for the heptahydrate , we find that the densities which he gives are much greater than those found by Lannoy , or by ourselves , the value of the density at 45 per cent , of the heptahydrate , according to Carhart , being 1'343 , instead of L318 , as given by our observations .
It is possible that the sample used in this case may have consisted largely of the hexahydrate , or the discrepancy may be due to other causes .
If we take older determinations of the density of zinc sulphate solutions , such as those of Gerlaoh or of Schiff ( 1859 ) , we find that they show a general agreement with our observations rather than with the formula of Bender , but that the characteristic point to which we have drawn attention is neatly smoothed out in the tables which they give as deduced from the results of their observations .
The point in question would not be noticed at all unless the observations were plotted by the method of differences , as shown in Fig. 3 , and even in that case it might readily be mistaken for an error of observation , unless the points were numerous , and had been inde144 Prof. II .
L. Callendar and Mr. H. T. Barnes .
pendently checked .
If we had ourselves foreseen a simple relation of this character , we should have taken even greater pains in verifying the observations about this point .
As it is , we hope shortly to be able to investigate the subject further , and in particular to endeavour to find similar relations in the case of other solutions .
Fig. 3 .
S 22 .
Change of E.M.F. with Strength of Solution at Constant Temperature .
A number of Clark cells of the B.O.T. pattern , but without crystals , were set up with the solutions above described , of which the density and composition had been carefully determined .
Due precautions were taken in each case to avoid evaporation .
The difference of E.M.F. from the standard at 15 ' C. , and also at other temperatures , was carefully determined in the case of each of these cells .
On plotting the results , we could not find any simple relation between the change of E.M.F. and the density or the percentage strength of the solutions .
But on expressing the observations in terms of w , the weight of ZnS04 per c.c. , and not per gram , of solution , we found that the values of cZE , the difference of E.M.F. from the standard at 15 ' C. , fell very nearly on a straight line , represented by the formula :\#151 ; dE = 42-0\#151 ; 88-0 ?
.{\gt ; ( millivolts ) .
The following table contains the observations for each solution tested :\#151 ; Electromotive Force of different Forms the Clark Cell .
145 Table III.\#151 ; Change of E.M.F. with Strength of Solution at 15 ' C. Type of cell .
( tv ) gram per c.c. ( ZE millivolts .
\lt ; ZE calculated .
Difference , obs .
\#151 ; calc .
r 0*105 332 32-8 + 0-4 B.O.T. 1 0-116 31 6 31 -8 -0-2 unsaturated ] 0-199 24-6 24-5 + 0-1 L 0-263 19 -1 18-9 + 0-2 Sat , at 0 ' C. . .
0 -401 6-5 6*7 -0-2 Sat. at 15 ' . .
0-478 o-o o-o o-o The differences given in the last column are of the same order as the accidental changes of E.M.F. observed in the case of these weaker cells .
It would therefore appear probable that in this type of cell the diminution of E.M.F. is simply proportional to the volume concentration of the salt .
In comparing the above results , a curious point remains to be noticed .
Taking a cell saturated at 15 ' C. , the increase of E.M.F. , on cooling down to 0 ' C. , has been shown above to be 8'4 millivolts , if there is no change of strength of the solution .
The increase of E.M.F. , due to change of strength of solution from saturation at 15 ' C. to saturation at 0 ' C. , has been found to be 6'5 millivolts .
We might , therefore , naturally expect the total effect due to both causes combined to be 14'9 millivolts , whereas the saturated crystal cells , in which both causes are operative , show an increase of E.M.F. of 16-6 millivolts .
The explanation of this apparent discrepancy is to be found probably in the lowering of E.M.F. , due to the greater diffusivity of the mei-curous sulphate in the weaker solutions .
In the saturated crystal cells this diffusion is practically prevented by the dense layer of crystals .
In order to test this hypothesis , some weaker cells were set up in the W form , in which the possibilities of diffusion are diminished by the smallness of the tube and the increased distance between the electrodes .
These cells showred , as was expected , higher values of the E.M.F. than those given by the formula , the difference amounting in some cases to between 2 and 3 millivolts .
S 23 .
On the Solubility of Zinc .
It is well known that zinc sulphate forms various hydrates , which may be obtained by crystallisation at different temperatures .
These hydrates differ in point of solubility , and it is important for Clark cells to employ the heptahydrate , which has the lowest solubility at Prof. H. L. Callendar and Mr. H. T. Barnes .
temperatures between 0 ' and 30 ' C. , and may therefore be considered the normal hydrate at these temperatures .
The peculiarities of the curve representing the temperature-variation of the E.M.F. of Clark cells are undoubtedly due in the main to the very considerable changes of solubility of this hydrate with rise of temperature , and are closely associated with the formation of different hydrates at higher temperatures .
We therefore thought that it would be of interest to investigate this point more closely , at least within the range of temperature over which accurate measurements of the E.M.F. may be readily obtained .
The solubility of zinc sulphate has been studied by Etard , * who discovered that the solubility of the sulphates decreased with rise of temperature at higher points of the scale .
His results are expressed in terms of the percentage weight , dissolved in 100 parts of the solution at a temperature t ' C. Expressed in this manner , he finds that the curve representing the solubility is a straight line , the equation of which for zinc sulphate is given as being p = 27-6 + 0-2604* ... ... ... ... ... .
( fitard .
) This equation is given as representing the solubility up to a temperature of about 80 ' C. , above which the solubility decreases .
Roscoe and Schorlemmer , f on the other hand , give the following table of solubility expressed in terms of the weight of ZnS04 dissolved in 100 parts by weight of water .
For the sake of comparison , we have reduced their results to the corresponding percentage , of solution , and have added two lines of results calculated from the formula of Etard , and from the observations of Poggiale .
Table IY.\#151 ; Solubility of Zinc Sulphate ( Roscoe and Schorlemmer , Etard and Poggiale ) .
Temperature O 0 ' 20 ' 50 ' 75 ' ZnS04 in 100 01 water ( R. \amp ; S. ) .
41-3 53-0 66 -9 80-4 Percentage of f R~ \amp ; S , 29 2 34 -6 40-1 44 -5 27 -6 32-8 40-6 47-1 Solution ( jp ) [ Poggiale 30-0 34 -7 40-7 45 -0 It is evident from the discrepancies shown in the last three lines that the matter requires further investigation .
From our own experiments we find that the rate of diffusion in these extremely dense and viscous solutions is so slow that it is much more difficult # ' Count .
Rend .
, ' vol. 106 ( 1888 ) , p. 206 .
f Yol .
2 , Part I , p. 262 .
Electromotive Force of different Forms of the Clark Cell .
147 to obtain correct results than might at first sight be expected .
In particular , we find it extremely important to maintain the solution for a considerable time with continual stirring at each point at a perfectly constant temperature .
This essential condition could be readily secured by the aid of the constant temperature baths already described .
The method which we adopted at each temperature was to prepare a saturated solution with repeated stirring in a lai-ge test-tube in the constant temperature bath .
This solution , mixed with crystals , was maintained for several hours at a constant temperature .
Samples of solution free from crystals were drawn off in a special pipette at intervals and weighed .
They were then evaporated to dryness at 100 ' C. , and the percentage of ZnS04 in each case was calculated , assuming the residue to be the monohydrate .
The different samples at each temperature always agreed very closely , showing the solution to have been saturated and free from crystals .
The results of our observations are contained in the following table .
With the exception of the last four lines , which represent single observations , each line is the mean of two or three determinations .
Observations in different lines were taken on different days and with different samples of solution .
Table V.\#151 ; Solubility of Zinc Sulphate ( Callendar and Barnes ) .
Temperature centigrade .
( * ) .
Percentage of solution .
O)- Calculated by formula 29 *5 4- 0 '270t .
Difference ( see Pig .
IV ) .
Obs.\#151 ; Calc .
0 ' 29 -43 29-50 -0-07 0 29-53 29-50 + 0-03 0 29 -49 29-50 -o-oi 15 -00 ' 33-66 33-55 + 0-11 15-88 33-85 33-78 + 0-07 30-70 38 -46 37-80 + 0-66 39-92 41 -36 40-28 + 1-08 39 -95 41-37* 40-29 + 1-08 40 -73 41-43 40-50 + 0-93 41-49 41-70 40 -70 + 1-00 46 -40 42-68 42 -02 + 0-66 49 -97 43 -51* 43 -00 + 0-51 49-99 43-41 43 -00 + 0 -41 50-00 43 -50* 43 -00 + 0-50 50-20 43 -51 43-05 + 0-46 The observations marked with an asterisk were obtained by heating the solution to a temperature ten or twenty degrees above that 01 the bath and then allowing it to cool down to the bath temperature with constant stirring .
VOL. lxti .
m Prof. H. L. Callendar and Mr. H. T. Barnes .
In order to exhibit the nature of these results on a convenient scale , it is necessary to plot , not the whole percentage , as is usually the case , but the small differences from the linear formula given in the fourth column .
It is then apparent that there is a singular point in the curve of solubility at a temperature of 39 ' C. At this point , as the temperature rises , the rapidly increasing solubility of the heptahydrate begins to exceed that of the hexahydrate , which is also increasing , but less rapidly .
At temperatures above 39 ' , a solution containing only crystals of the heptahydrate becomes supersaturated with respect to the hexahydrate , so that if any crystals of the latter are formed or introduced , there will be rapid crystallisation , and the strength of the solution will diminish .
At 39 ' C. the solubilities of the two hydrates are equal and amount to 41'1 per cent , of the solution ; but whereas the rate of increase of the solubility of the heptahydrate is 0-33 per cent , per 1 ' C. , that of the hexahydrate is only 022 per cent , per 1 ' .
The crosses marked R. and S. in fig. 4 , at the points 0 ' and 20 ' , correspond to the values given by Roscoe and Schorlemmer plotted on the same scale .
The remaining observations of Roscoe and Schorlemmer , and also those of Etard , differ so much from ours that Fig. 4 .
/ \#166 ; \#163 ; / .O .
-6 \lt ; H\gt ; ?
'a * O Ccvrre of D/ f ) Cerence \gt ; \lt ; r from / / j Unedi - For/ m. Za . !
\#187 ; .
\ ^X p =\#163 ; 9 5 + \#166 ; \#163 ; ?\#163 ; / Z P \ , i \gt ; PdOA S\ .X \gt ; \#151 ; ; 0 ' 2 O0 6 O0 4 0 ' 6 0 ' 6 j\#163 ; t Ll J they could not be included in the limits of the page .
The observations of Poggiale do not show the diminution of solubility above 80 ' C. which Etard gives .
There is a general agreement in the results quoted , especially between those of Roscoe and Schorlemmer and Poggiale , but if the differences may be taken as indication of the order of accuracy attained , it is evident that the results of these observers could not be expected to show the critical point at 39 ' C. By interpolation on the curve given in fig. 4 , we find the following values for the solubility as given by our observations .
Electromotive Force of different Forms of the Clark 149 Temperature C. ... ... 0 ' 10 ' 20 ' 30 ' 40 ' 50 ' Per cent , of sol .
( p ) . .
29-50 32-24 35-13 38 -22 41-33 43-45 ZnS04 in 100 aq 41 -85 47-58 54-16 61 -86 70-44 76 -84 Compare Poggiale* ... 43-02 48-36 53 -13 58 -40 63-52 68 -75 Difference -1-17 -0-78 + 1-03 + 3-46 + 6-92 + 8-09 S 24 .
Change of E.M.F. at Higher Temperatures .
( 30 ' and 50 ' C. ) Finding that there was a change in the continuity of the curve representing the temperature variation of the E.M.F. at temperatures above 41 ' C. , we decided to investigate this point more closely , although the temperatures in question lie beyond the range of the practical use of Clark cells .
By a slight modification of the heating arrangements , the constant temperature baths were enabled to reach steady temperatures up to and beyond 50 ' C. The following series of observations were taken with several cells of different types , in a manner similar to that which has been already described .
Each line represents the mean of the different cells at each point .
The observations given in different lines were taken on different days , in the order in which they are given in the table .
An interval of several months intervened between the first and the last half of the table .
Table VI.\#151 ; Change of E.M.F. between 30 ' and 50 ' C. Temperature centigrade .
Difference in millivolts from 15 ' C. Difference from linear formula ( L ) .
Observed .
Formula ( L ) .
40 -60 ' 35-81 30-72 5-09 30-14 19 -78 18-17 1-61 30-32 20 -00 18-38 1 -62 35-44 27 *39 24-53 2 -86 42-58 36-56 33-10 3-46 46-74 41 -14 38 -09 3-05 48-58 43-50 40-30 3 -20 35-79 28-08 24 -95 3 13 40 09 34-99 30-11 4-88 42 -79 36-76 33-35 3-41 44-70 38-88 35-64 3-24 41 -54 35-45 31 -85 3-60 * ' Watt 's Dictionary of Chemistry .
' Muir and Morley , vol. 4 , p. 581 .
M 2 Prof. H. L. Callendai and Mr. H. T. Barnes .
The differences from the linear formula given in the last column of Table VI are shown by the crosses on the curve in fig. 5 .
This figure is plotted , as in the previous case , to exhibit not the whole change of the E.M.F. , but only the defect of the change from lineality .
For instance , at 39 ' C. , the E.M.F. does not suddenly begin to increase , but continues to diminish at a slower rate .
It was evident on plotting these observations that there was a break in the curve , and that the observations after 41 ' C. belonged to a different branch .
To investigate this point more closely , a continuous set of readings was taken on a pair of very sensitive cells .
Starting at a temperature of 35 ' C. , the temperature of the bath was Fig. 5 .
Scaieof Temperature Centigrade .
50 ' *0 ' 50* r r r Difference of \#163 ; .
d.F in Miiiiisoite From Lint ?
are Formu/ a , rh , , i 1 1 , .
X ' ''^4 .
very slowly and continuously raised by disconnecting the regulator , and readings were taken of the temperature and E.M.F. of the cells alternately every few minutes .
The E.M.F , of the cells was observed to fall at a steady rate , accurately following the curve already found for the observations at steady temperatures , until a temperature of 42'4 ' was reached .
At this point there was a sudden increase of more than 2 millivolts in the E.M.F. , due to rapid crystallisation of the hexahydrate , and in less than five minutes the cells had reached a point on the other branch of the curve .
After remaining at this point for some hours , the bath was slowly and continuously cooled , observations being taken in the same manner during the cooling .
As had been expected , the cells were found to follow the branch of the curve shown in fig. 5 , corresponding to the solubility of the hexahydrate .
The solution remained supersaturated with respect to the heptabydrate until a temperature of 31 ' was reached , at which point the E.M.F. had fallen nearly five millivolts below the normal .
At this point there was a sudden crystallisation of the heptahydrate , and the E.M.F. rose in a few minutes to its normal value .
It will be observed that the two branches of the curve Electromotive Force of different Forms of the Clark Cell .
151 cross at 38-8 ' , which is in practical agreement with the temperature of equal solubility of the two hydrates , as determined by the observations on the change of solubility .
S 25 .
Permanence and Reproducibility of B.O.T. Crystal Cells .
It is now more than a year since the completion of the experiments described in the preceding pages , and we are able to add the results of more recent comparisons of the cells as evidence of the order of their permanence and reproducibility .
Of the original crystal cells , made more than two years ago , we still have a few remaining .
None of the cells of this type have shown any signs of failure , in spite of the treatment to which they have been subjected , but many of them have been taken away by the students who made them .
Table VII.\#151 ; Comparisons of Crystal Cells at 15 ' C. Differences from mean of cells in millivolts .
of cell .
1 Dec. 10,1895 .
Nov. 28,1896 .
Dec. 19,1896 .
Feb. 8 , 1897 .
Aug. 2,1897 .
X 1 -0-03 -0-04 -0-08 -0-02 -0-04 X 2 -0-05 + 0-02 + 0-06 -0-04 + 0-02 X 3 + 0-10 + 0-04 + 0-05 . .
+ 0-12 .
X 5 + 0-03 \#151 ; 0 '08 -0-07 + 0-02 \#151 ; '-12lg f \lt ; N X 6 + 0-08 + 0-07 4-0-08 + 0-04 X 10 -0'07 + 0 -04 + 0-01 . .
+ 0-10J -2 X 11 -0-08 -0-06 -0-07 . .
-0-02 ^ As there is no particular reason why these cells should be less permanent than other Clark cells of the B.O.T. type , the above will probably be sufficient proof of permanence .
It will be observed that the average difference from the mean in each case is nearly one-twentieth of a millivolt .
The extreme difference is one-tenth .
We have observed that this is about the order of agreement generally attainable with Clark cells set up at different times .
Over comparatively short intervals of time , such as one month , it would appear from the above list , and from other tests , that the average change in the value of any one cell , as compared with the mean , may be expected not to exceed two or three hundredths of a millivolt , but for longer periods , such as a year , the mean change reaches one-twentieth .
As a further illustration of the reproducibility of these cells , and of the close agreement in the temperature variation of the E.M.F. , under somewhat exacting conditions of testing , we add a list of the 152 Electromotive Force of different Forms of the Clark Cell .
cells made by the fourth-year class of electrical students during the session of 1896-1897 .
These cells were all of the portable B.O.T. form , with a flattened and amalgamated platinum wire in place of mercury .
They were set up in test-tubes , filled with crystals , and sealed with marine glue , and were otherwise exactly similar to the crystal cells described in S 10 .
With the exception of those marked with an asterisk , the observations were all taken by the students themselves in the course of an afternoon 's work .
Readings were taken at the points 15 ' , 0 ' , 15 ' , 80 ' , 15 ' , allowing only about half an hour at each temperature .
Table VIII.\#151 ; Portable Crystal Cells , made by Students .
Name of student .
Date when made .
Date when tested .
Difference from standard at 15 ' .
Change of E.M.F. Mean change per degree .
0 ' \#151 ; 15 ' .
15'-30 ' .
Stovel Mar. 2 . .
Mar. 4 . .
-0-17 16-57 19 -39* 1 -199 Thomson Mar. 2 . .
Mar. 4 .
* -0-11 16-60 19-38 1 -199 Blair Feb. 19 .
Feb. 21 .
+ 0-05 16-63 19 41 1-201 Burnham Dec. 8 . .
Jan. 19 .
-0-02 16-57 19 -30* 1196 Davidson Jan. 12. .
Jan. 19 .
+ 0-11 16-62 19 -40 1 -201 Edwards Dec. 10 .
Jan. 26 - + 0-07 16 -60* 19 -36* 1-199 Macbean Dec. 3 . .
Dec. 8 . .
-0-04 16-62 19-36 1-199 McDonald , P. W. Jan. 14 Jan. 19 .
+ 0-14 16-63 19 -49 1-204 McDonald , J. E. Feb. 19 .
Feb. 21 .
-0-15 16-58 19*30 1-196 Packard Jan. 12 Jan.19 .
+ 0-14 16-56 19'43 1 -200 Pitcher Dec. 10 .
Jan. 26 .
+ 0-05 16 " 58* 19*53 1-204 Walters Dec. 3 .
Dec. 8 .
-0-13 16-61 19 -44 1-202 Mean of students ' cells -0-005 16 -603 19-400 1 -2000 It should be observed that the cells were , in most cases , tested rather too soon after being sealed up .
In the course of a week or two , they were usually found to have settled down into closer agreement with the standard .
When kept for a longer time than half an hour at 30 ' C. , they showed a slightly greater change of E.M.F. at this point .

