Standard Resistance-coils with Mercury Standards. 351 



and it was clear that there was some small change progressing 

 in the tube. It was carefully examined, but no trace of an 

 air-bubble could be seen ; and then it was emptied, cleaned 

 again, and refilled, with the results given in Table I. Much 

 the same was observed with coil No. 39. No. 38 was cleaned 

 and filled twice, but gave perfectly consistent results all through. 

 This uncertainty seems to me to constitute one serious objec- 

 tion to the general employment of mercury tubes as standards. 



I believe that for the observations recorded in the above 

 Tables the tubes were properly filled, and that there were no 

 bubbles of air in them. This, I think, is shown by the agree- 

 ment between the results of different fillings, and I suppose 

 that the resistance will not seriously alter so long as the same 

 mercury remains in the tubes ; but it appears that, after fill- 

 ing, electrical experiments are required to make certain that 

 every bubble of air has been removed, and that the tube 

 really has its true resistance. I should imagine, too, that the 

 tubes would require somewhat frequent refilling to make sure 

 that the mercury may remain pure. My experiments showed 

 that it was almost impossible to keep the inside of the cups 

 above the mercury perfectly dry. It was necessary, w x hen 

 making the comparisons, to remove the glass stopper and 

 insert the contact-pieces; and this had to be done when the 

 tube was immersed in the ice-water. 



The cups were open to the air for a short time in making 

 this change, and that short time was quite sufficient to cause 

 a deposit of dew to be formed on the inside of the cup. So 

 long as this slight moisture remained in the cup and did not 

 reach the tube itself, of course it did not affect the results ; 

 but it would be difficult to feel certain that after a time the 

 mercury in the tube was quite dry. 



This difficulty would be avoided by working at the tempe- 

 rature of the room rather than at that of melting ice. The 

 large temperature-coefficient of mercury, from three to four 

 times that of platinum-silver alloy, is, however, an objection 

 to this. 



Another difficulty, caused by the necessity of working 

 at zero, was that the mercury in the cups was always 

 slightly warmer, 0°'2 or 0°'3 C, than the ice. This was, no 

 doubt, caused by the conduction of heat down the copper con- 

 necting-rods, and from the upper portions of the glass of the 

 cups which were exposed to the air. This, of course, would 

 necessitate a small correction to the values of the resistances 

 given in the Tables; but the correction must be exceedingly 

 small, for the temperature of the mercury in the tube itself, 

 where it is actually in contact with the ice, must be zero, 



