242 
Proceedings of the Royal Society of Edinburgh. [Sess. 
the same figure for conductivities, would be quite possible for waters which 
are taken as having the same chlorine value. It will be seen from Table 
III. that this difference is frequently doubled ; but, as stated above, errors 
in measurement of the conductivity would account for a further difference 
of 2 in the fifth figure, and the error in density determinations would 
account for a difference of the same order of magnitude ; so that, as a whole, 
the conclusion is reached that either density or conductivity measurements 
give a chlorine value sufficiently near the true one for all practical purposes 
in oceanography. 
This may be shown by taking the density or conductivity values farthest 
from the mean for any given chlorine values and calculating the chlorine 
value corresponding to the density or conductivity from Table IV. Thus 
in Nos. 46 and 60 we get the relative densities A 18 » = 1 *02779 and 1*02766. 
These differ from the mean density for all the waters of the same salinity by 
+ 0 00008 and — 0*00005. These differences correspond to differences in 
chlorine of 0 06 gr. and 0 04 gr. respectively. If we allow a titration 
error in these extreme cases of 0 02 grams, it will be evident that the 
approximation to accuracy is quite sufficient when the values corresponding 
to the densities are taken in place of the titration values, the actual errors; 
being about 0*04 gr. and 0'02 gr. in the two extreme cases, or less than 0*2; 
and 01 per cent, of the total chlorine. 
Taken as a whole, the values obtained from the conductivities are slightly- 
less accurate, the average variations being a little more than double those- 
of the densities ; while for 0T00 of chlorine the difference in density should! 
be (H)0013 and in conductivity 0*00024, or slightly less than double. In 
the most extreme case, No. 56, the variation from the mean is —0*00018,. 
which corresponds to a difference in chlorine of —0*075. The admitted 
possible error is 0*02 (titration) + 0*01 (for specific conductivity) = ± 0 03. 
Thus the difference between the value from the corrected conductivity 
would differ from a corrected titration value by 0*045 grams of chlorine in 
the most extreme case. This, again, is little over 0*2 per cent. 
It is apparent at once that the angles of minimum deviation obtained 
above are of little value in comparison with the other methods. This is, 
probably due to the fact that the instrument used was not sufficiently 
delicate. It would be necessary to use a scale reading to at least 5" of 
arc to obtain the same degree of accuracy as is given by the other 
physical methods. As stated above, that used by the author read 
directly to 30". The optical measurements are, in consequence, not 
further discussed. 
The conclusion that may be drawn from these experiments is that certain 
