.S7A 1 U'/LLIAM .SYAM/A'.V.V, /-\K.S. 175 



c. Although, by careful selection, two tubes of nearly equal 

 diamrtrr may be obtained, it would not be safe to depend upon 

 Biicli uniformity where accurate results are required. Each tube 

 should, then-fore, be calibrated, and provided with its own scale; 

 ami, in case of a tube having to be replaced, a suitable new scale 

 should also be provided. The smaller the diameter and the greater 

 tin- length of the tubes, the greater will be the accuracy of the 

 "1'StTvations ; but a limit is here imposed, by the necessity of the 

 gas-bubbles rising freely to the surface, which limit is reached in 

 m luring the tubes to 2 millimetres of diameter. 



A much smaller diameter would suffice, if the gases were 

 merely to propel a water-column before them in a horizontal tube, 

 but I found that under such circumstances the resistance of the 

 liquid by adhesion to the sides, caused considerable error and 

 inconvenience in the manipulation of the instrument. 



Having measured numerous resistances by this instrument, and 

 compared the results with measurements obtained by a very perfect 

 "Wheatstone bridge arrangement, I find that it may be relied upon 

 within one-half per cent, of error of observation, excepting at the 

 extremes of the range, where a somewhat greater amount of error 

 easily occurs unless special care be taken in reading the compa- 

 ratively few divisions on the one side. A higher degree of accuracy 

 is, in such a case, to be attained by filling the one tube several 

 times (noting the volume each time), and allowing the other to 

 continue accumulating, until at least 100 divisions of the scale 

 shall have been passed. 



A table has been prepared which gives the temperatures cor- 

 responding to the volumes of the gases of decomposition observed 

 in the tubes, thus saving all calculation on the part of the metal- 

 lurgist, or other observer.* 



* The manner in which the equation of the curve of increase of resistance with 

 lemj>erature is applied to the construction of the table here referred to is the 

 following : the coefficients of the platinum wire employed, that is, the quantities 

 o. , 7, have first to be calculated, from a series of experiments made for that 

 l'iirijse, with one unit of resistance at zero Centigrade. The constant of the 

 voltameter y has next to be obtained in the manner explained at p. 172, and the 

 resistance X of equation, p. 172, has then to be equated with that of r, given at 

 p. 148. 



The following is the calculation employed for the construction of the tables. The 

 constant is equal to 17 units, the resistance y to 2 unite, the platinum coil in 

 the pyrometer has a resistance of 10 units at zero Centigrade, and the coefficients 

 of the platinum wire employed are a = '039309, $ = '00216407, y = - '24127 ;. 



