390 The Absolute Value of the Mechanical Equivalent of , terms of the International Electrical Units .
By Howard T. Barnes , D.Sc .
, Macdonald Professor of Physics , McGill University , Montreal .
( Communicated by Prof. H. L. Callendar , M.A. , LL. D. , F.R.S. Received April 30 , \#151 ; Read May 27 , 1909 .
) In 1902* there were published the results of an extended series of measurements of the capacity for heat of water at different temperatures between 0 ' and 100 ' C. The continuous method of calorimetry was employed in the work , the details of which were developed by Prof. H. L. Callendar and myself .
A steady electric current was used to heat a steady flow of water through a fine bore tube , bringing about steady temperature conditions in the calorimeter .
The energy given in joules per calorie was expressed in terms of the international electric units , and was consequently dependent on the values of the standards used in the measurements .
The very greatest possible care was exercised in selecting the standards employed and in making sure that they represented the values assumed for the international electrical units .
Previous to the commencement of the investigation of the capacity for heat of water , several years were devoted to a thorough study of the Clark cell as a standard of electromotive forcef and to the determination of the temperature coefficient of the Clark cell .
From a large number of cells suitable examples were selected which were as perfect as could then be obtained .
Additional cells were also made at different times and with different materials , in order that those actually used in the investigation could be expressed in terms of the mean of as large a number as possible .
In addition , a further investigation !
of the Weston cell was made , and the mean of all the Clark cells was compared with the mean of all the Weston cells .
The value of this ratio was found to be Clark at 15 ' C. Weston at 20 ' C. = 1-40666 , which is practically coincident with the value of the ratio found by Jaeger * ' Phil. Trans. , ' A , vol. 199 , p. 149 ( 1902 ) .
+ ' Roy .
Soc. Proc. , ' vol. 62 , p. 148 ( 1897 ) ; B. A. Report , Toronto , 1897 ; ' Electrician , ' vol. 39 , p. 68 ( 1897 ) , vol. 40 , p. 165 ( 1897 ) .
I ' Journ. Pliys .
Chem. , ' vol. 4 , p. 339 ( 1900 ) , vol. 8 , p. 196 ( 1904 ) .
The Absolute Value of the Mechanical Equivalent of Heat .
391 and Lindeck* for the cells prepared at that time at the Reichsanstalt .
They found Clark at 15 ' C. Weston at 20 ' C. = 1-40665 .
While entire confidence was felt that we had cells which represented the true value of such combinations , the uncertainty still existed as to the absolute value of the cells .
In view of this fact , and until such time as future work might throw light on the question , we adopted the legal value in calculation .
This is given as 1*4342 int .
volts at 15 ' C. Recent investigationsf have shown that this value of the Clark cell is too high .
The most probable absolute figure , so far as it is possible to fix it , is now given as 1-4330 int .
volts at 15 ' C. for the cells set up according to the new specifications .
There is , however , an important difference between the cells set up according to the old specifications , such as were used in our determinations of the mechanical equivalent of heat , and the cells set up according to the approved methods now adopted at the various standardising laboratories .
The fundamental difference lies in the treatment of the mercurous sulphate .
By the old specifications this was washed with water before , being mixed with the zinc sulphate to form the paste .
According to the newer methods , no water is allowed to come in contact with the mercurous salt , which is specially made by various methods , all of which give results in very good agreement .
F. A. Wolff and Waters !
have shown that the cells set up according to the old specification are 0'30 millivolt higher than the new cells .
With the assistance of Dr. H. L. Bronson we have recently equipped the laboratory with a set of modern standards , some of which were made by Dr. Bronson in Washington , and directly compared with the cells at the Bureau of Standards .
Other cells have been made here by him with the assistance of Mr. A. N. Shaw , all of which have agreed very closely with the cells constructed in Washington .
Details of the comparisons of these cells will be published shortly .
It has been possible for us to construct several cells in an exactly similar way to those previously used for the mechanical equivalent of heat measurements , and to compare them with * ' Zeit .
fur Instk .
, ' vol. 21 , p. 33 ( 1901 ) , vol. 21 , p. 65 ( 1901 ) .
t K. E. Guthe , 'Bull .
Bureau Standards Washington , ' vol. 2 , p. 33 ( 1906 ) ; T. Mather and F. E. Smith , 'Koy .
Soc. Proc. , ' A , vol. 80 , p. 383 ( 1908 ) .
X ' Bull .
Bureau of Standards , ' vol. 4 , p. 64 ( 1907 ) .
Prof. H. T. Barnes .
Absolute [ Apr. 30 , our new cells .
The result of this test has shown that the mean of our cells , set up according to the old specifications , as our old cells were constructed , are 0*28 millivolt higher than the modern Clarks .
As a matter of fact , our new cells are exactly 18 microvolts ( 0*018 millivolt ) higher than the reference standards at Washington , which brings the mean of our cells set up , according to the old specifications , to 0*30 millivolt .
We have also set up a number of inverted Weston cells , using the washed mercurous sulphate , and find them approximately the same amount higher than the modern Weston cells prepared according to the new specifications .
The absolute value which we must therefore take for our cells in order to reduce the values of the mechanical equivalent must be 0*30 millivolt higher than the value assumed for the modern Clarks , or 1*4330 + 0*0003 = 1*4333 int .
volts at 15 ' C. It is possible now to express the original determinations on the basis of the new value for the Clark cell .
The value assumed for the resistance standards used in the investigation must be also assumed .
The actual resistance through which the heating current was passed was compared with 11 standards , all of which had certificates signed by Dr. Glazebrook , and a twelfth standard from the Reichsanstalt .
All these standards were expressed in terms of the international ohm and agreed with each other to better than 1 part in 10,000 , except one which was subsequently found to be in error and which was not included in the mean .
It is exceedingly unlikely , therefore , that either the mean of the resistance standards or of the Clark cells were in error by as much as 1 part in 10,000 from the international units , whatever values may ultimately be assigned to these in C.G.S. units .
In the calorimetry also great pains were taken to eliminate errors of thermometry and heat loss , and the agreement of the observations , taken at different times and under widely different conditions , was not far from 1 part in 10,000 .
The method of water jacketing rendered the calorimeter independent of external temperatures .
This made the work at the extremes of the temperature range as accurate as those at ordinary temperatures .
It is difficult to see where the measurements can be in error by more than 1 part in 10,000 at any part of the range .
In the table following is given the values of the thermal capacity of water at different temperatures between 5 ' and 95 ' C. in joules per calorie .
Column 1 contains the original values expressed in terms of the Clark cell equal to 1*4342 int .
volts at 15 ' C. Column 2 contains the same results reduced to the modern Clark equal to 1*4330 int .
volts at 15 ' C. , or what is the same value for the old Clarks , 1*4333 volts .
In Column 3 is given the 1909 .
] Value of the Mechanical Equivalent of Heat .
values of Kowland by the direct mechanical method obtained between 5 ' and 35 ' C. as corrected by Waidner and Mallory .
Table I. Temperature .
Joules per calorie Clark 1 *4342 .
Joules per calorie Clark 1 -4330 .
Rowland Mechanical .
5 4 *2105 4 -2052 4-206 10 4 *1979 4 -1926 4T96 .15 4 -1895 4 -1842 4T88 20 4 -1838 4 -1785 4T81 25 4 *1801 4 -1748 4-176 30 4 *1780 4 -1727 4-174 35 4 *1773 4 -1720 4-175 40 4-1773 4 -1720 45 4 *1782 4 -1729 50 4-1798 4 -1745 55 4 -1819 4 *1766 60 4 -1845 4 *1792 65 4 -1870 4 -1817 70 4 -1898 4 -1845 75 4 *1925 4 -1872 80 4 -1954 4 -1901 85 4 *1982 4 -1929 90 4 -2010 4 -1957 95 4 -2038 1 4 -1985 Means 4 -1888 4*1835 The means of the two columns are seen to be , for the interval between 5 ' and 95 ' C.\#151 ; For Clark ... . .
T4342 4T888 joules per calorie Clark ... . .
1-4330 4*1835 It is a matter of interest to compare these values with the direct mechanical measurements of Rowland , and of Reynolds and Moorby .
On account of the limited range of Rowland 's experiments comparison can only be made between 5 ' and 35 ' C. The mean value over this range from Rowland 's curve is found to be\#151 ; Rowland 's mean value ... ... ... 4*185 joules .
From the second column of results , the mean value over the same range is found to be\#151 ; Electrical mean ( Clark 1*4330 ) ... ... 4*1826 joules .
Here a divergence of 1 part in 2000 is shown between the two values , but when we enquire into the limits of accuracy which Rowland assigns for the values he gives , we find that this is about 1 part in 1000 .
Hence we find an 394 Prof. H. T. Barnes .
Absolute [ Apr. 30 , agreement of 1 part in 2000 , which is better than Rowland gives as the probable accuracy of his measurements .
When we come to the absolute measurements of Reynolds and Moorby , we find probably greater accuracy on account of the method they adopted .
They used the continuous method with steady temperatures and heated the water from the freezing to the boiling point .
The value they obtained was the mean over the entire range , and was found to be\#151 ; Reynolds ' and Moorby 's mean ... ... . .
4T832 joules .
Prof. A. W. Smith* has recently pointed out that their results ate really for an interval of temperature between 1 ' and 100 ' C. To express this mean value accurately between 0 ' and 100 ' C. , it becomes necessary to increase their value slightly , since the value of " J " increases rapidly between 1 ' and 0 ' C. Assuming the variation of the specific heat of water from our determinations , this value is increased to 4T836 joules for the mean calorie .
Over the same interval our mean value must be increased slightly , making allowance for the increase at both ends of the temperature range .
This brings the value 4T835 between 5 ' and 95 ' C. up to 4T849 between 0 ' and 100 ' C. , which is in accord with their value to 1 part in 4000 .
The value is therefore seen to be between that of Rowland and that of Reynolds and Moorby .
The excellent agreement shown here indicates that we must , at the present time , have a knowledge of the electrical units not far from the truth , both as regards the Clark cell and resistance .
The agreement with Rowland to within the limits of accuracy of his measurements serves to link his absolute mechanical measurements with those of Reynolds and Moorby in a very satisfactory way , and to show that they obtained almost identical values .
The value of the 15 ' C. calorie is a matter of considerable importance\#151 ; Rowland gives this as ... .
4T88 joules Our value comes ... ... ... .
4T84 " Assuming the variation of the specific heat , Reynolds ' and Moorby 's value comes\#151 ; 4183 .
* 'Monthly Weather Review , ' vol. 35 , p. 458 ( 1907 ) 1909.1 Value of the Mechanical Equivalent of Heat .
A paper has recently appeared* in which the 15 ' C. calorie has been discussed .
Our value of this quantity is given as\#151 ; 4*187 joules .
This value is given as expressed in terms of a Clark cell equal to 1*4334 int .
volts at 15 ' C. The measurements for the Clark cell here quoted were carried out by an absolute method in the Macdonald Physics Building by Mr. R. O. King and described by Prof. Callendar , f but were only preliminary and no record has been kept of the particular cell used .
Although the value is remarkably close to the correct figure for our old cells , it can hardly be taken as representing the mean Clark cell which was .adopted in the calculations for the mechanical equivalent .
Mr. King .eventually intended to obtain more accurate readings for our Clark cell mean , but was unfortunately called off on other work .
The work is being .continued in this laboratory after a lapse of several years , and it is hoped that further results may be communicated later .
[ Note by Prof. H. L. Callendar.\#151 ; I designed the apparatus used by Mr. R. O. King to read to an order of accuracy of 1 in 100,000 , and personally assisted him in all the fundamental measurements .
I had no .doubt at the time that the result which I worked out from Mr. King 's observations , though not so perfect as might have been obtained in a more extended investigation , correctly represented to 1 in 10,000 the absolute value of the Clark cells at that time employed in our investigations in terms of the international ohm .
It is very satisfactory to find that Dr. Barnes ' recent comparisons agree so well with the old absolute measurement , but I think the absolute value of the mechanical equivalent deduced from the electrical standards may possibly be in error by at least 1 part in 4000 .
] * * " Verh .
der Deut .
Phys. Ges .
, ' vol. 6 , p. 589 ( 1908 ) .
t 'Phil .
Trans. , ' A , vol. 199 , p. 55 ( 1902 ) .

