68 REroKT 1888. 



then -99969 at 15-5, showing a fall of -00031, No. 34 was 1-00014., show- 

 ing a rise of -00014. Since this date No. 19 has fallen therefore by -0008, 

 and No. 34 by -00051, and these numbers are within the limits of error 

 of the fall of "00065 found for the platinum silver coils. We would infer 

 then that while apparently there was a serious change in these coils rela- 

 tively to the platinum silver standards between the date of issue and 1875, 

 since that date there has been no change. On referring to Mr. Taylor's 

 letter on p. 66 it will be noticed that the history of these coils previous 

 to 1875 is uncertain ; all that is known is that they have the foi'mal B.A. 

 stamp, and it is stated in the Fourth Report of the Committee, 1866, that 

 all the coils issued are correct to "0001 at the temperatures stated. 



There is still another coil of some interest. This is now marked ^^ 54. 

 It was made in accordance with the suggestion of Chrystal in 1876, with 

 a thermoelectric junction attached. Fleming compared it with his 

 standards in 1879 and 1880. In 1884 it Avas again compared by us and 

 found to have the value "99658 B. A.U. at 8"3, with a temperature coefficient 

 of -000295. It was then sent to Professor Kohlrausch at "Wiirzburg for 

 comparison with some mercury units constructed by Strecker, and was 

 returned by him at the end of his experiments. 



In 1888 it was again compared and found to be "99653 B.A.U. at 8"3, 

 ■with a coefficient of -000290. It will be seen that the change is "00005, 

 which is within the temperature errors. 



Thus we conclude, from this general account of the condition of the 

 coils at present, that with the exception of the platinum iridium coils A 

 and B there is no evidence of ar!y change of as much as "0001 B.A.U. since 

 the years 1874 or 1876, but that all the platinum silver coils and the two 

 gold silver coils belonging to Mr. Taylor changed apparently by about 

 "0007 B.A.U. between the time of their construction and the time at -which 

 they were examined by Chrystal and by Taylor respectively. This change 

 may of course be a real one ; we incline, however, to suppose that it is 

 apparent only, and offer the following explanation, already several times 

 referred to. 



Hockin says in a note to his Table of Temperatures, ' British Associa- 

 tion Report,' 1867, which gives the temperatures for the standard coils of 

 the Association : ' The values given in the above table are deduced from 

 the german-silver coil called B ^ used in your Committee's experiments 

 in 1864.' 



He does not seem to have compared among themselves the standards 

 of various materials, but to have referred each to B. Now we are ignorant 

 of the temperature at which the comparison was made, but we know he 

 used the coefficient "00032. This at present is too high by "00004. If 

 we suppose that Hockin made his determinations with the coils in ice, 

 then this error in the temperature coefficient would lead him to a valuB 

 for the coil at 15°, which would be too high by -0006. 



Having once got a platinum silver coil supposed to be known, it would 

 be natural to use it as a standard rather than any of the others, because 

 of its low temperature coefficient, and the error made in the original 

 determination of G would thus be pei-petuated. This conclusion is borne 

 out by the observations on Messrs. Elliott's coil No. 41, Table X. Its 

 standard temperature fell apparently by 2° between the time of its issue 

 by Matthiessen in 1864 and Hockin's comparison in 1879, and then rose 



' This is not the same as our B. 



