MEASUREMENTS OF ELECTROLYTIC MIGRATION 13 



was analyzed, " a" denoting the anode compartment, " c" the 

 cathode compartment, and " ac" meaning that both were analyzed 

 and used in computing the transport number; next comes the 

 transport number of the anion, here the cipher before the decimal 

 point is always omitted, and the number is filled out to three 

 decimal places by dashes when the experiments were less ac- 

 curate ; attached to the transport number is given half the 

 maximum divergence (multiplied by i,ooo) between the indi- 

 vidual results, preceded by the sign ±. 



Thus " 23 ex a .170 ±4" means that 23 experiments were 

 performed, the analyses at the anode compartment were used 

 to compute the transport, the transport number found for the 

 anion was 0.170, the maximum divergence between the indi- 

 vidual results was 0.008. Very inaccurate experiments would 

 be given as follows : " 5 ex ac .51 — ± 67 " meaning that the 

 transport number found is 0.51, and the difference between the 

 two extreme results was 0.134. Noyes' results are given thus : 

 ** ac .1657 d= 2.3 " meaning 0.1657 with a maximum difference 

 between the results of 0.0046. I am convinced that this maxi- 

 mum divergence does not in most cases give an exaggerated 

 impression of the inaccuracy of the experiments. 



The arrangement just described is followed as far as possi- 

 ble ; often, however, a column is left out when the data are 

 lacking, or when they have been given in the paragraph at the 

 top. Extra columns, special arrangements of the data, etc., 

 are explained in each special case. 



In deciding what experiments it was fair to average together, 

 in almosf all cases I have been much more tender with the 

 results than the authors themselves ; many authors show what 

 they think of their own results by averaging together all their 

 values, even when obtained under different conditions of tem- 

 perature and concentration. When only one experiment was 

 performed under each set of experimental conditions, I did not 

 feel justified in averaging the results even where their accuracy 

 seemed questionable. Thus I have tried to give the experi- 

 ments the benefit of the doubt ; the reader should take this into 

 account when trying to decide in any particular case whether 

 the transport numbers really vary with the temperature, etc., 

 or not. 



