On the Determination of Chemical Affinity. 283 



values obtained by Moser as the E.M.F.'s set up by the inter- 

 diffusion of solutions of zinc sulphate or of copper sulphate 

 respectively, with the same limits of strength of solution. 



It seemed of interest to pursue this point further, so as to 

 decide by accurate measurements how far this approximation 

 is merely an accidental coincidence : accordingly a large 

 number of experiments have been made, mostly described in 

 the following pages. 



At first the observations were made by means of the quad- 

 rant electrometer, no permanent current passing ; but it was 

 found that the experimental and observational errors were 

 such that unless a very large number of duplicate cells were 

 prepared and very many observations made with each, in- 

 volving an immense expenditure of labour and time, trust- 

 worthy averages could not be obtained. The electrometer- 

 readings (deflections on both sides of zero) could, at best, be 

 read to no nearer than + # 0025 volt; so that averages for 

 single-cell values with a probable error of not more than + *001 

 volt (one millivolt) entailed a large number of readings for each 

 cell ; and as the values for different cells thus obtained exhi- 

 bited a range of several millivolts, numerous duplicate cells 

 had to be set up to obtain a final mean with a probable error 

 of less than ±1 millivolt as the E.M.F. of each kind of cell: 

 whilst, finally, the differences between the mean E.M.F.'s 

 thus obtained exhibited, when plotted, deviations from a regular 

 mean curve approaching to + 1 millivolt, even with the utmost 

 care in making observations and the bestowal of much time 

 in so doing. Accordingly, the method of observation was 

 changed, a delicate galvanometer being substituted for the 

 electrometer ; the instrumental errors being smaller and the 

 sensitiveness greater*, the same degree of accuracy (i. e. the 



* The galvanometer (by Elliott) was capable of being adjusted so as 

 to give a total deflection, reading on both sides of zero with a reversing- 

 key, of upwards of 200 scale- divisions when the current from a Daniell 

 cell was passed through a resistance of 100,000 ohms, using the T -oVo" shunt ; 



1-114 

 so that one scale-division = i ooo y 1 00 000 v 900 ampere= "000056 micro- 

 ampt3xt3. 



It being convenient to have some term representing the millionth of a 

 millionth (-000000000001=10- 12 ), it is proposed to use the prefix " elasso " 

 for this purpose (from eXacro-ov), just as " micro " represents 10 -6 , and 

 " mega" represents 10 6 : the value of 1 scale-division as above then becomes 

 56 elasso-amperes. The fraction -000000001 = 10 ~ 9 may be conveniently 

 represented by the double prefix " milli-niicro," using each half in its 

 ordinary sense ; so that a centimetre = 1 milli-micro earth-quadrant. In 

 the same kind of way 1,000,000,000 ohms may be termed a Hlomegohm-, 

 and a million of million ergs may be spoken of as a meizerg (/ielfoi/). This 

 nomenclature seems to be practically more convenient than such terms as 

 twelfth-ampere, ohm-nine, erg-twelve, &c. 



U2 



