MR. 11. M. DICKSON ON THE CIRCULATION OF THE 
These seemed so consistent, and as the work progressed the chance of a serious error 
failing to be caught on the charts seemed so small, that duplicate determinations 
were afterwards made of onl}^ every fourth or fifth sample, merely to make sure that 
one’s eye was not “ out” in deciding on the end-points. 
The standardising of the silver solutions with sufficient accuracy gave some httle 
trouble. The silver nitrate was obtained from Messrs. Johxsox, Matthey & Co. 
as triple crystallised and fused, and notwithstanding very careful testing in various 
ways, no impurity of significant importance was detected, and in particular no free 
acid. The solution was made up in Cjuantities of about 5 litres at a time, and each 
lot was stored in two Winchesters and treated, so far as standardising went, as two 
separate solutions. The strength of each solution was ascertained by titrating 
'weighed cpiantities (about 10 grammes) of four different solutions, containing approxi¬ 
mately the same amount of chlorine as an average sea-water ; two of these solutions 
were prepared from carefully purified KCl, and two from similar NaCl, and it was 
arranged that the supply of more than one solution should not run down at the same 
time, so that each fresh solution should always control, and be controlled by, three 
old ones. The following S 2 )ecimens, selected at random, indicate the general accuracy 
obtained - 
Silver solution No. 24 — 
Gramme Cl in i cub. centim. 
KCl solution A . 
. . . -00692.3 
,, ,, B . . 
. 
. . . -006936 
NaCn ., C . . 
. 
. . . -006919 
„ a. 
• 
. . . -006930 
i\Iean . 
. . . -006927 
Silver solution No. 65- — 
KCl solution A. 
. . . -007-fit 7 
,, ,, B . . 
. . . -007259 
NaCd „ C . . 
. . . -007245 
,, 1>. . 
. . . -007252 
Mean . 
. . . -007256 
These values are consistent to vrithin '00002 gramme, again well within the limits 
recpiired. 
In the earlier 2 :)art of the work, the conpxirison with some of the Siu'engel tube 
determinations, to be descril)ed presently, raised some doubt in my mind as to the 
real accuracy of the jnethod, and the reason which first suggested itself was the quite 
distinct difterence in the appearance of the end reaction with KCl and NaCl solutions 
and with sea-vrater, the last being sharjDer and showing less tendencv to ‘‘ go back.” 
Ibis seemed to indicate an uncertainty (lue to a varying jiersonal equatifiii ni judging 
