DEPOSITED BY THE FACULTY OF GRADUATE STUDIES AND RESEARCH
t,'i:/^...-J- • -;*/
MSGILL UNIVERSITY LIBRARY
ACC. NO. DATE
laniTENAKGE OF PRECISION STAIffiARDS OF ELECTROMOTIVE FORCE
b7 A. Henrikson, B, A*
Being a thesis submitted to the Faculty of Graduate Studies and Research, McGill University, in partial fulfillment of the requirements for the degree of Master of Science•
Ibntreal April, 19U8, ACIQTOWLEDGEriENT
The writer wishes to express his thanks to Dr. A. Norman Shaw for his very helpful suggestions and guidance during the progress of this inve s ti gation.
To Dr. G. ¥.Vinal of the National Bureau of Standards in Y/ash- ington, D. C, he owes a debt of gratitude for his kind cooperation in making possible the international comparisons and for providing historical information concerning the development of these measure ments .
He also wishes to thank Dr. Haaer of the Bureau of Standards for his Idnd help in making the actual measurements in Washington.
iii TABLE OF CONTENTS Page 1» Introduction (a) Historical 1 (b) Object of the Investigation h (c) Definitions h (d) Theory of the Ageing Equation 5 2. Apparatus and Procedure 7 3. Experimental Results and Conclusions (a) Recapture of the Initial Reference Mean 13 (b) NeiT Cell Comparisons mth the Bureau of Standards in Yfeshington, January, 19U8. 20 (c) General Conclusions 23 U. Appendices (a) Bibliography 2k (b) Notes on the Restoration of Old Cells 2^
iv SUMARY
Coii^arisons v/ith the National Bureau of Standards at Tfashing- ton, D. C. in Januaiy, 19U8, have indicated that a reference mean of electromotive force can be recaptured even after a period of forty years to within twenty microvolts. This has been accomplished with unexpectedly high precision Isy application of a previously establish ed ageing correction to the mean electromotive force of a small number of old cells. Confirmation is provided, of the validity of the ageing correction "vrfiich is of the form: Ej, = A + B'log (M -I-T), where A, B, and T are constants, and M is the age of the reference cells.
il Section 1
INTRODUCTION
(a) Historical
The standards laboratories of the world have long been concern ed with the maintenance of precision standards of voltage, current and resistance. In lQ9h} the International Congress adopted the Clark cell, subject to certain specifications, as the standard of electro motive force. Its value was defined as l.IOU volts at 15 C.
Later it was shown that the Weston cell was superior in several vrays to the Clark, and in I908 the London conference adopted the
Weston cell as standard, defining its value in terms of the internat ional \inits of resistance and current. In 1910 a special technical committee representing the national physical laboratories of England,
France, Germany and the United States met in Washington to make conip- arisons of st andards of resistance, current and voltage. This group constructed a large number of cells in Yfashington and defined the international volt in terms of the mean emf of these cells. It recom mended that the mean value of these cells be taken as I.OI83O volts at 20*-' C. and this was adopted by the International Committee in
October, 1910, and became effective in Januaiy, 1911. These cells were then divided up among the participating countries to be used as the primary standard of voltage for their national physical laboratories.
However, the value of the volt has not remained the same in all countries, due partly to the differing methods for maintaining the unit.
In the United States the basis of reference is now a group of cells
"vrfiich have proved to remain very constant for long periods of time. No account is taken of ageing which is assumed to be negligible, and
changes in the mean voltage of the group are made only when cells are
added to or removed from the batch. In the National Physical Labor
atory in England new cells are constructed each year and the reference mean is the average of the cells made in the preceding three years.
In 1923 the United States reference basis was about ten microvolts
lo;Ter than the English basis and at the present time is about eighteen microvolts lower.
In 1908, Bronson and Shaw, at the suggestion of Dr. H. T, Barnes,
set out to equip McGill University with a set of modem standards.-^
A number of standard cells were constructed, some of them in Y/ashing-
ton, to the new specifications of Wolff and Waters. The average value of these cells was found to be in close agreement with the means
assumed in London and Washington at that time. Tests carried out
during, ;the summer of 1912 and during the winter of I9I8 - 19 indicated that the initial reference mean could be recaptured to within t\TO parts in 100,000 if a total ageing correction of about sixty microvolts was applied to cells ten years old.^ Many exchanges with London and Imsh- ington Yfere made during this time.
In order to discover more about the ageing of these cells work vras carried on at McGill under the direction of Dr. A. N. Shaw. In
1923, R. J. Clark constructed more than fifty normal Weston cells, some
1. Bronson, H. L. and Shavr, A. N., The Electrician, Vol. 6S. pp. 698- 702, (1911).
2. Wolff, F. A. and Vfaters, C. E., Bull. Bur. Stand. Vol. U, No. 1, Reprint No. 70 (1907).
3. ShaY7, A. N. and Reilley, H. E., Trans. Roy. Soc. Can. Vol. 13. pp. 171 - 176. (1919). ' with acid and some with neutral electrolyte. The ageing of these cells was observed by means of further comparisons with London and Washington.
Shaw showed that the ageing of the neutral cells could be represented
very accurately by means of the semi-empirical relation:
Ej^ = A + B'log (UfV), where A, B, and T are constants and M is the age of the cells, while the acid cells remained practically constant
over a period of years .-^ This made it possible to recapture the inter
national mean of voltage iirith much better agreement T/ith the Y/ashington
and London means than had been possible heretofore. A further paper
on the maintenance of a standard of electromotive force was published by Shavf and Reilley outlining the procedure to be followed in setting up a standard of voltage and in recapt\iring an initial reference mean.^
1. Shaw, A. N., Reilley, H. E., and Clark, R. J., Phil. Trans. Roy. Soc London, Series A, Vol. 229, pp. 125 - l62. (1935TI 2. Shaw, A. N., and Reilley, H. E., Can. Jour, of Research, 3: pp. U73 - U89. (1930). ^ (b) Object of the Investigation
The object of this investigation was (l) to recondition the McGill' standard cell laboratory, (2) to test the collection of old cells vjhich had been previously compared ivith those at London and at Washing ton, (3) to test the ageing correction for Weston normal cells over a period of forty years and (U) to determine the accuracy with which the international reference mean for voltage could be recaptured through out a long period lay a small collection of cells, and ydthout necessarily maiding nevT cells or new comparisons with other standard laboratories.
(c) Definitions
(i) The "ageing" of standard cells refers to the decrease of electromotive force ivith time.
(ii) The Weston "normal" standard cell is that primary source of electromotive force constructed according to the specifications of
Wolff and Y/'aters, (loc. cit.), in which the electrolyte is a saturated solution of cadmium sulphate.
(iii) The "international mean" of emf was defined in 1911 in terms of the voltage of the Weston normal cell. The mean of a large number of cells constructed by the International Congress in 1910 and 1911 was defined as I.OI83O international volts.
(iv) The "initial reference mean" is that voltage which is chosen by a laboratory as its primary reference. In this investigation the initial reference mean was the mean voltage of number of cells which, when measured in 19115 agreed to Td.thin several microvolts with the newly defined international mean. (d) Theory of the Ageing Equation In the development of the semi-empirical relation giving the ageing of the Weston cells, Shaw, Reilley and Clark, (loc. cit.) have discussed several possible causes of ageing, and the conclusion is that the most probable cause is the slow solution of foreign ingredients from the glass container. The amounts of these solutes are very minute and approach a limit asymptotically. Thus the emf of a cell may be represented by an equation of the form: e = a'+- b' lo^c where c is the concentration of the solute. If, at construction the amount of the impurity x is equal to a small quantity b, and E = EQ , then E = EQ -^ B loge| The rate of accumulation of x ivill be —; = k(a-x) where a is the saturation value dt' of X, or the total amount available. Therefore x- a(l- e'^^') If t = T when X = b, and we measure t from this time i.e., t - t - XT and b = a( 1 - e"^^ ) then E ^ EQ t- B log^-^j-^^^ ^ EQ ^ -;°-^g-^)(;--1?^-v_-)] Since k is very small and '^ is usually less than 5? this simplifies to: 5V = E^ -K B log, t^ ^.p \ + B log, X or E^ *:= E^^^ir B log^{ t i-T ) and subtraction of the initial reference mean from both sides Ej^ ^ A + B log (M +r ) Because of the small size of Z and k, the simpler approximation will hold for the useful age of the cells. Ej^ is the difference from the
initial reference mean at the age of M months.
If a number of cells are averaged, the ageing equation becomes:
B^n A +• B log^( M+tr), where ^, A are the arithmetic means, and ^ may be taken as either the arithmetic or the geometric mean if T is nearly the same for all the cells or if M is large.
B has been found to have the value -17.1 for all normal cells,
or -39.U if common logarithms are used. The average value of T for all cells made in 1923 was 0.75. Section 2 APPARATUS AI€) PROCSDURS The apparatus required for the making of precision measurements of the electromotive force of the cells was essentially that set up in 1908 by Shaw for his early investigations.-^ For the removal of uncertainties due to temperature, a thermostatically controlled oil bath was provided. The temperature of the oil was controlled by four elements through which the current vfas adjusted by means of a bank of four carbon filament lamps in parallel. Tvro of these lamps were controlled manually to supply most of the heat lost, and were adjusted according to the room temperature. The other two were controlled automatically to keep the temperature of the bath constant at 25.0 Centigrade. A schematic diagram is given in figure 1.
Pttentic^fntt-er
FIG. 1 Electrical Connections to the Constant Temperature Bath
1. Bronson, H. L. and Shaw, A. N., Brit. Assoc. Advancement Sci. Repts. 79: ^9e - 398. (1909). ~ ' The cells to be measured were set up in racks in this bath as shown in figure 2, and connections were made to three parallel lines above the bath from which connections were made to the potentiometer.
FIG. 2
Showing the Method of Mounting the Cells in the Bath
To keep the tenrperature of the entire bath uniform, the oil was kept in motion by means of a propeller and shaft mounted in the center of the tub and run by a D. C. motor.
The thermoregulator, figure 3) consisted of about four and one-half meters of glass tubing, about ij centimeters diameter, bent into the form of a grid, figure U, and filled ^wvith toluene. 7
FIG. 3
Upright Portion of the Thermoregulator
^^^^^^^B/^^^f^^^^^^^^^*^^^^^^^^j^^
FIG. I|
Toluene Rilled Tubing Fonning the Thermoregidator 10
The end of this grid was bent down to form an elbo?r and then into
an upright. The purpose of the elbow is to prevent the toluene from
rising into the upright portion which was filled ivith mercury. The
top end of this part was joined to a capillary tube into which the
mercury was pushed by the expansion of the toluene. A platinum wire
sealed into the upright belcrw the capillary formed one terminal vfhile
a second contact was made by means of a wire inserted dovm the open
end of the tube. Yi/hen the mercury made contact i^ith this upper ter
minal, a rel^ vras activated and the controlled laups shut off, thus
allomng the bath to cool slightly.
A considerable amount of time was spent reconditioning the thermo stat and the rest of the apparatus. The reconditioning of the thermo stat was particularly difficult as the glass had become extremely brittle iTith age.
It was found that this thermostat limited the temperature vari ations usually to within 0.01 degree, but on occasion when the room temperature was low this variation reached 0.02 degree. However, since the temperature coefficient of Yfeston normal cells is only about one- half microvolt per hundredth degree, this would not introduce appreci able error.
The temperature of the bath and the changes of temperature vieve measured with a calibrated Beckman/ithermometer reading to l/lOO degree.
Since all cells were at the same regulated temperature, differences in
1. ShaiT, A. N., Trans. Roy. Soc. Can. Series 3, Vol. 11, pp. 133 - 13U. (1917). ^^ n
electromotive force between cells should be the same for any constant temperature over a range of at least several degrees.
In mailing measurements, one cell was chosen as a reference and all other cells were measured relative to it. This was done by oppos ing the tY/o cells and measuring the difference of voltage on a single slide-wire potentiometer along which was a potential drop of one micro volt per millimeter. Thus the differences could be read directly from the millimeter scale placed below the wire. Since the differences were in most cases small an error of even 5^ in the current flowing in the slide wire would not affect the results appreciably. Two reversing switches were provided, one to reverse the cell connections, and the other to reverse the battery terminals for the potentiometer.
This eliminated the need for corrections for thermoelectric effects and contact potentials.
FIG. 5
Thomson l^e Galvanometer Used in Cell Measurements 12
The galvanometer, figure $, was an old Thomson reflecting type.
Its sensitivity was such that Yfhen placed at a distance of about ten feet from the scale, it gave a deflection of about one scale division per microvolt. HcriTever, when tr.TO cells were opposed in series the internal resistance of the cells cut do^m this sensitivity by about one-half or more.
It was found at first that touching any part of the bench or woodvTork surrounding the potentiometer would cause a large deflection in the galvanometer. j\lso the readings obtained on the cells were not consistent and totally different results were obtained by reversing the switches mentioned previously. It was discovered that sufficient current was leaking through the woodwork from the D. C. povrer lines
feeding the stirring motor and the heating coils, to cause such an action. It is important to note that when dealing with extremely small currents as ±n this experiment that the leaks from the D. C. povrer lines across ordinary insulators may have an appreciable effect on the ultimate microvolt reading of the potentiometer. This effect v;-as completely eliminated when the entire system was insulated. All con necting Tdres were mounted on good porcelain insulators and the poten tiometer system, including the galvanometer, vr&s moimted on glass sheets. 13
Section 3
EXPERBffllTTAL RESULTS AND CONCLUSIONS
(a) Recapture of the Initial Reference Mean
In order to recapture an initial reference mean, a record of the emf of the cells at an early period of their existence must be avail able. From these records the constants in the ageing equation may be determined and the correction obtained for any subsequent date. The routine for doing this is given in ShaiT's paper, (loc. cit.). All of the cells used in the reference group had been made during the first
week of September, 1923. Table gives the readings of these cells on several later dates. The readings for those spaces left blanlc are not available.
TABLE 1
RECORD OF THE AGEING OF THE REFERENCE CELLS
Date Sept. 20, Oct. liay, May, Dec. Aug. 1923. 1923. 192U. 1925. 1926. 1928. Age 0.5 month 2 months 8 months 20 months UO months 00 months Cell ^ 60 t33 + 19 - 2 -31 -li3 -71; 62 •^33 t20 - 2 - 9 -58 63 -^36 + 20 ^ 7 + 3 -13 -29
6k 'i-39 i 20 f 8 + 3 + 8 + 7 6$ -t38 •^20 -3 -U8 66 +32 + 20 + 3 -10 -2U -3S 61 4-27 +17 + i|. - 3 -lU -11; 69 fllii •^^3 ^Sk + 22 - 3 -1h 71 ^$$ •^k3 + 15 -26 -21 -1^0 79 ^kl - 1 -21; -28 -31 -36 Means + 38 +20 fk -12 -21 -30 lu
In obtaining the average for each date above, the selection rule
stated by Shaw has been applied; viz., reject any cell whose deviation
from the mean emf of the batch is greater than 20 -H d, where d is the mean deviation.
In this investigation all cells in the reference group were the
same age, but this is not necessary although it does simplify the calculations. If they are not all the same age, a correction must be applied to the older cells in order to bring them into line with the rest. For example, if several cells five years old are to be aver aged with some which are ten years old, one only needs to determine how much the cells would age betiveen the fifth and the tenth years.
It is possible to do this since the slope of the ageing curve is the same for all neutral cells. This value is then added to the emf of the ten-year-old cells, thus bringing their value in step rdth that of the newer cells.
Figure 6 shows the ageing curve dravm for the cells of table 1.
The difference from the initial reference mean is plotted against
M + r, on semilogarithmic paper. The value of V is talcen as 0.75 as stated on page 6. This value is verified by the straight line form of the graph. This graph may be produced to predict the mean value of the cells on any subsequent date, or the ageing correction may be cal culated from the equation of the straight line:
E^l ~ hl.5 - 39»k log^^(M t 0.75). This gives an average correction for these cells at the present time, age 290 months, of --SS microvolts.
1. Shaw, A. ,N. and Reilley, H. E., Can Joum. of Research, 3: i;73- U89. (1930). 15
'BK '
^V ^ ^h^ 5 « 5 8 « "»N .c % ^ m t « 9 m '0 «C >0 ^ V «» C •»i /• C t *<< •;/*«•* •'W U 5. ? • 1 1 ? 9 * ^-^ * ,, .««4y «JM«i»/f*Cf 16 Table 2 includes the averages of a number of readings of the cells as referred to cell 63. The next columns indicate the procedure in recapturing the initial reference mean using the ageing correction determined above. The selection rule here vfould eliminate cells 6k and 71. However, the mean of the remaining cells is still 2 microvolts referred to number 63. These cells are then expressed in terms of their mean by subtracting 2 microvolts from each reading and in terms of the initial mean by adding the ageing correction, "$6 microvolts. Thus the mean of these cells is now 56 microvolts below the initial reference mean. TABLE 2 ILLUSTPuATBIG THE RECAPTURE OF THE INITIAL REFERENCE Iv5EAN Cell No. Referred to Deviation Referred to Referred to from mean mean of group initial mean 60 -23 25 -25 -81 62 -k 6 - 6 -62 63 0 2 -2 -58 6k ^k3 la •hia -15 6$ - 3 5 - 5 -61 66 -12 U; -lU -70 61 + 31 29 -f-29 -27 69 ^-15 13 1-13 -k3 71 -kS 1;7 -U7 -103 79 -^19 17 •M7 -39 Mean •f 2 - - -S6 Mean deviation 20 — - 17 Table 3 expresses the voltage of the remaining cells in terms of the initial mean, as obtained from the reference group of table 2. In the last column is given the predicted values for the mean of the acid cells, and for the mean of the old cells. It thus offers confir mation of the validity of the ageing equation of the newer cells and indicates the usefulness of the equation even up to forty years. TABLE 3 OTHER CELLS EXPRESSED IN TERMS OF THE BIITIAL liEAl^I Cell No. Referred to Referred to Mean Predicted #63 initial mean mean values [83 -7 -6S Acids 86 + 122 ^6k X -71 -71 I 91 -19 -77 /C2 •»-32 -26 19091 1.6 +17 -Ul -^3 -kl (if z ^ 0.75) cellsl^l.8 -3U -92 X - omitted in averaging because of discordant values and erratic behaviour in the past. Shaw has shown that acid cells remain relatively constant for a number of years, Tdth perhaps a very slight decrease of electromotive force, and that the initial value is 66 microvolts below the reference mean. From Shaw's results this would become about -71 microvolts in tvTenty-five years, that is, at the present time. This agrees ivith table 3 which gives the mean of cells 83 and 91 as -71 microvolts. 1. Shaw, Reilley and Clark (loc. cit.) 18 Thus an assumption of -71 microvolts from previous Tfork as the mean of the acid cells, would have resulted in the same ageing correction for the neutral cells as that determined by the ageing equation. However, the average of only two cells is not in general acceptable, but it is illuminating to consider several previous values of cell 83. In table 1; it is seen that its present value as determined by the ageing of the neutral cells is consistent ip/ith its previous record. This seems to indicate that the acid cells that do survive for many years retain their remarkable constancy at least up to twenty-five years, TABLE k RECORD OF ACID CELL #83 May, 192U Nov. 1925 Oct. 1928 Dec. 19U7 -70 ^66 -69 -65 The second group of cells in table 3} which are now forty years old, also illustrate the value of the ageing equation. These cells were all made in September,• 1909^ but no records of their values are available until 1912. Thus it is impossible to obtain an estimate of the value of V in the ageing equation. However, an approximate equation can be obtained from the data in table 5 which gives records of these cells on three vd-dely separated dates. TABLE 5 AVERAGE OF CELLS C2, 1.6, 1.8 REFERRED TO INITIAL JilSAlT Date July, 1912 May, 1919 Aug. 1925 Dec. 19U7 predicted Age 31; months 116 months 191 months value. -13 -19 -26 -kl 19 If C is assumed to have the value 0.75, substitution of the above values in the ageing equation gives an average value for A as 57.8. Thus the ageing equation for these cells is: E^ ^ 57.8 - 39.1; logjM -^0,75) This gives an ageing correction for December, 19l;7j of -1;7 microvolts. This is the most probable value of the correction for these cells. If V" had a value as high as 25, the highest observed value, the ageing equation would become: 1»^ "=^62.9 - 39.1; log,^(M +^25), giving a correction of -1;3 microvolts. Thus an error of only U microvolts would be made by neglecting f • It is therefore seen that the use of these old cells to determine the international mean would produce an error of possibly 10 microvolts as compared Tri.th the estimate based on the neirTer neutral and acid cells. These cells had a very high internal resistance which made it difficult to obtain an exact balance on the potentiometer. There is a probable error of '^^ microvolts in these measurements. Thus even mth only three cells forty years old, it is possible to recapture the international mean of voltage to within 20 microvolts. If a larger number of cells v^ere available with complete records of their early behaviour, the international mean could probably be recaptured to within a fe?r microvolts. 20 (b) New Cell Comparisons Tdth the Bureau of Standards in Yfeshington, January, 19l;8. In order to test the results of the ageing equations used in the previous section, and the approximate methods based on comparison of acid and neutral cells, six cells were carried to the National Bureau of Standards in 'iTasliington, D. C. in January, 19l;8. These cells were placed in a temperature regulated oil bath similar in principle to the one described in this paper, and kept at 20 degrees Centigrade for several days. The emf of these cells v^ere measured on a Leeds and Northrup Standard Cell Comparator, which measures difference from a reference cell but is made direct reading by setting on its dials the voltage of the reference cell^ The average of four series of measure ments made in Washington by the author and Dr. Kamer of the Bureau of Standards is given in table 6. TABLE 6 VALUE OF CELLS MEASURED IN WASHINGTON Cell No. anf as measured on Ihf referred to c omp arator. (volts) international mean (microvolts) 62 1.018282 -18 63 1.018275 -25 6$ 1.018269 -31 66 1.018256 -10; 69 1.018261; -36 83 1.0l82l;2 -58 These cells were brought back to McGill University and remeasured using cell 60 as a reference. Several other cells which had been left 21 at McGill, vrere also measured to ascertain cell 60 had remained cons tant. The results of these measurements are foimd in tables 7 and 8. TABLE 7 TEST OF CONSTANCY OF #60 IN DEC. AND J Ml. Cell No, Dec. 19l;7 Jan. 19l;8 Change 52 -lU -lU 0 61; •^66 + 62 -3 67 + Sk t5o -1; 71 -21 -26 -5 79 + ll2 ^36 -6 Average change -I; Table 7 indicates that the average of five cells has decreased by 1; microvolts relative to #60, or which is more probable, cell 60 is now four microvolts higher than when the original measurements were taken. Thus this cell is noT/ -77 microvolts referred to the initial McGill reference mean as indicated in table 8. TABLE 8 SUiaWJlY OF THE UTTSRNATIONAL COMPARISONS Cell No. Referred to Referred to Referred to Difference McGill mean McGill mean Bureau mean between Wash, before trip after trip in Wash. and McGill 60 -81 -77 - - 62 -62 -1;7 -18 -29 63 -58 -56 -25 -31 6S -61 -51 -31 -20 66 -70 -57 -10; -13 69 -1;3 -1;5 -36 - 9 83 -66 -68 -58 -10 22 Table 8 gives the results of the measurements on the cells after their return from Washington. It also includes for comparison, the voltage of the cells referred to the initial McGill reference mean before the trip and the values in terms of the Bureau of Standards reference mean. In view of the fact that these cells received a bad shaking on the joum^ to Washington but were kept relatively steac3y on the return trip, it is thought advisable to consider only the third and fourth columns of table 8. Furthermore, since the values of the cells rela tive to one another are not the same in ?/"ashington and at McGill, it is felt that reliance can be placed only in those readings wliich remain ed practically unchanged before and after the trip, if at the same time the value of the cells relative to one another is the same both in the Washington and McGill measurements. The last three cells fit these requirements. The difference between the Bureau values for these cells and the McGill measurements indicates a difference of 11 microvolts in the reference means, the McGill mean being the higher. If the average of the six cells is used then the McGill reference mean is 19 microvolts higher than that used by the National Bureau of Standards. However, it is believed that the first figure is closer to the actual value. 23 (c) General Conclusions. The use of the ageing equation has thus made it possible to recapture the international mean of voltage possibly to mthin eleven microvolts of the Vfashington standards and certainly to within twenty microvolts. The value of McGill reference mean is found to lie betvTeen the United States' and the British reference means. The ageing equation has been verified over a period of forty years by con^arison with newer neutral and acid cells and also by direct international comparisons. 21; Section k APPEIIDICSS (a) Bibliography Bronson, H. L. and Shaw, A. N., The Electrician, Vol. 66, I9II. Wolff, F. A. and Waters, C. E., Bull. Bur. Stand. Vol. k, No. 1, 1907. Shaw, A. N. and Reilley, H. E., Trans. Roy. Soc. Can. Vol. 13, 1919. Shaw, A. N., Reilley, H. E.,and Clark, R. J., Phil. Trans. Roy. Soc., Series A, Vol. 229, 1930. Shaw, A. N., and Reilley, H. E., Can. Jour, of Research, Vol. 3, 1930. Bronson, H. L. and Shaw, A. N., Brit. Assoc. Advancement Sci. Reports. 79, I909. ShavT, A. N., Trans. Roy. Soc. Can., Series 3, Vol. 11, 1917. 25 (b) Notes on the Restoration of Old Cells One of the commonest faults of old cells is their tendency to develop high internal resistance. This may be due to two conditions, (i) the formation of gas between the cadmium amalgam and the cadmium sulphate crystals in the negative limb, or (ii) the hardening of the crystals after having been pushed up by the expansion of the amalgam. This second cause would come about if the temperature had been raised higher than usual and slov^ly lowered. In some cases these cells may be restored by heating to about 30 or 35 degrees Centrigrade accompanied by gentle shaking. This loosens the crystals and brings them into contact ivith the amalgam. Another method is to open one limb of the cell and tamp do\7n the crystals mth a clean glass rod. The cell is then resealed. This was tried for a number of old cells whose high resistance made their measurement impossible. The voltage of these cells were observed soon after being opened and thereafter for several weeks. Initially the emf rose to several hundred microvolts above the initial mean but dropped rapidly. Figure 7 shows that the drop is very nearly of a logarithmic form but the slope is much steeper than the ordinary ageing curve. After several weeks the cell had dropped to values that were consistent with its age. This method of restoration would be advisable only as a last resort if a standard cell were needed badly when only old high resis tance cells vrere available. 26 5 e M. %f "^ ;s 4 * I K *• i^ s < "> 1 1 « « ? ? ? I t ' • » McGILL UNIVERSITY LIBRARY uriACC.