50 REPORTS ON THE STATE OF SCIENCE. 



The mean difference in the values of all the manganin coils for 1908 

 and 1903 is 12-6 parts in 100,000. The oldest coils are 381 (seventeen 

 years old), 780, 738, 739, and 740 (thirteen years old), and 1693 and 

 1690 (eight years old), the ages being approximate only. The remainder 

 of the coils are from five to six years old. 



The most constant coils belonging to various groups are : — 



.S81 — most constant of the unit coils. 

 738 „ „ 10 ohms coils. 



730 „ „ 100 



710 „ „ 1.000 „ 



In general, therefore, the older the coil the more constant does it 

 appear to be. 



With reference to the sudden changes in resistance, as shown by the 

 curve for 2351 in 1903, of 381 in 190-1-5, and of L-20 in 1906, we can 

 offer no complete explanation : but it is possible that variable humidity of 

 the surrounding medium, such as might arise from the presence of a 

 small quantity of moisture in the insulating oil, was responsible for part 

 of these changes. 



The breaks in the curves for 2183, 2351, L-17, L-18,_1693, and 2484 

 are due to these coils being away from the National Physical Laboratory ; 

 they were being compared with the wire standards of the Reichsanstalt. 



The increase in resistance of No. 2449 is phenomenal. The daily 

 rate of change for 1906 is over four parts in 10,000,000 ; that is, in 

 about twenty-two days the coil changed in resistance by about one part 

 in 100,000. In April 1907 we attempted to measure the change fi-om 

 day to day, and for this purpose we compared 2449 and 740 every 

 working day for four weeks. The results obtained are as follows : — • 



Day of Observation . 

 Difference (2449-740) 



Parts in 100,000 

 Day of Observation . 

 Difference (2449-740) 



Parts in 100,000 



The change was, therefore, a very gradual one, and easily detected. 

 It is of interest to note that the rate of change for the last six months of 

 1907 is less than that for 1903-6. 



The possible causes of the changes in the manganin resistances may 

 be classified under the following heads : — 



1. Change in structure of the alloy. 



2. Surface action. 



3. Humidity effect. 



4. Change in the soldei'ed joints connecting the wires of high- 



resistance coils to the current leads. 



5. Change at the junctions of the potential leads with the 



resistance standard. 



Only the first of these appears to fully explain the gradual rise in 

 i-esistance. Causes 2 and 4 would have an inappreciable effect on very 

 low resistances ; yet some of these — e.g. 2196 — have changed by con- 

 siderable amounts. Cause No. 5 would have no effect on high-resistance 

 coils, since these are not provided with potential leads ; but Table XIII. 

 shows that all of the high-resistance coils have changed. Cause No. 3 

 produces in general a cyclic change, and, while being without doubt a 



