74 
MR. II. N. DICKSON ON THE CIRCULATION OF THE 
gave me perfect confidence in the chromate method for the purpose I had in viev\ 
The results seem to show conclusively that without spending the time necessary for 
the highest degree of accuracy, but merely taking ordinary care, the chlorines of 
large numbers of samples can be determined with comparatively little trouble, to 
within the admissible limit of error of i '03. 
In working out the chlorine titrations and computing the corresponding salinities, 
I have employed tallies of four-place logarithms, and have in no case gone beyond the 
second place of decimals in the results. The use of the third figure, which cannot be 
significant, is misleading, and its absence greatly simplifies the calculations. 
Although I was unable to use the material collected for any special investigations 
into the properties of sea-water, I thought it desirable to determine the specific 
gravities of a certain number of each hatch of samples, partly to get further evidence 
of the accuracy of the factors used in calculating the salinities, partly to act as a 
check on the chlorine work, and partly to give warning of any impurity in the samples. 
Six samples, usually the two saltest, the two freshest, and two others, were, as a rule, 
taken from each box, after the chlorine estimations were finished, and their specific 
gravities determined by means of Spreiigel tubes containing about 50 cub. centims. 
The determination was made at 15° 0., the tubes lieing set in a frame immersed in a 
tank, in which water wa,s kept actively circulating by means of paddles, driven by a 
small motor. The thermometer used was a standard, made and tested by Casella, 
and repeatedly verified by myself in comparison with standards in use in the 
laboratory. The tubes were left in the tank for ten to twenty minutes after any 
change of volume was apparent, and set before the paddle was stopped—the whole 
time in the bath being thirty to fifty minutes. They were then carefully dried and 
placed in an open cardboai'd box beside the balance for some time before weighing. 
The general accuracy of the pykiiometer method, even when sj^ecial precautions are 
not taken to get the best possible results, may be judged from the following, in Avhich 
the tubes were filled with the same samples 1 and 2 :— 
Sample. 
Tube. 
1. 
2. 
2 re-weighed next day. 
1 
1026-17 
1026-22 
1026-24 
2 
•12 
•26 
•21 
3 
•18 
•19 
•19 
4 
•13 
•23 
•19 
5 
— 
•28 
•29 
6 
— 
•20 
to 
o 
the second 
sample the first weigh: 
inof was made durino; 
weather of a cyclone front, and the second under the drier conditions succeeding. 
The following is a specimen of du})licate determinations taken in the ordinary 
course of routine work : — 
