320 Proceedings of Royal Society of Edinburgh. [sess. 
might be due to alterations in the chemical composition of a number 
of the samples combined with the effect of evaporation, due to 
imperfect stoppering on others. Such an explanation would have 
the effect of invalidating, to a great extent, the labours of these 
eminent observers, but I am glad to say that it appears to me 
that this apprehension is unfounded. 
Dittmar’s estimate of the constant hydrometer error is based upon 
the assumption that his estimate of the value for D was applicable 
to all the samples which he examined. It must, however, be 
remembered that he determined this value for D not in a number of 
different samples, but in a mixture made from a number of different 
samples. Obviously if the value for D varied appreciably in samples 
from different localities, the result of his experimental determinations 
of the value for D would depend upon the selection which he made. 
Although there is, unfortunately, little or no information available on 
this point. I am pretty certain that his value was distinctly too high 
for many of the samples which he examined, and very much too low 
for others. Be this as it may, his method of reasoning undoubtedly 
led him to underestimate the importance of one part of his work. 
He made a direct experimental comparison between the results of 
his plunger method for the determination of specific gravity and 
those obtained by Buchanan’s hydrometer method, and in making 
this comparison he used the identical hydrometer with which 
Buchanan worked on .board the “ Challenger.” Hov/, Mr H. 
Dickson has recently stated, in the paper to which I have already 
referred, that he found a constant difference between the values 
for qSq as determined by my method and those obtained by using 
a “ Challenger” hydrometer, and reducing the observed value 48^ to 
qSq by means of Dittmar’s tables — the use of which tables he more- 
over shows introduces no appreciable error. Further, as Mr Dickson 
points out, the amount of this constant difference was identical with 
that observed by Dittmar in comparing his plunger method with 
Buchanan’s hydrometer method. 
Thus the result of direct experimental comparison between 
Buchanan’s hydrometer method and two gravimetric methods, 
differing greatly from each other, was that Buchanan’s hydrometers 
gave, in the case of waters having a density approximating to 
1028 , a result which is too low by 0-12. 
