THE EXPANSION OF SEA-WATER BY HEAT. 
419 
Having thus linked together the results of our observations under one general law, it 
remains to compare our conclusions with those of Prof. Hubbard, whose investigation 
is, as we have already stated, the only one -which covers the same ground as our own. 
If, as Mr. Buchanan' points out, the specific gravities of different specimens of sea- 
water vary only between l - 02 7 8 and 1'0240, the results to be of value must be correct 
to the fourth decimal place ; obviously therefore the formulae of reduction ought to be 
as accurate as the experiments themselves. 
In Maury’s work Professor Hubbard’s Table of the thermal dilatation of sea-water 
extends from 22° F. to 200° F., the volume of water at 60° F. being taken as unity. In 
order to compare it with our own, we have transcribed the portion contained between 
22°F.( — 5°-5 C.) and 95°F. (36° C.), making the volume at 0° the unit, as in our own Table. 
Table XVI. 
°0. 
Hubbard. 
T. & R. 
°0. 
Hubbard. 
T. & R. 
— 5-55 
1-00012 
+ 2-0 
1-00009 
1-00014 
5-00 
1-00006 
40 
25 
33 
4-44 
1-00003 
6-0 
45 
56 
3*89 
1-00000 
8-0 
71 
83 
3-33 
0-99998 
10-0 
98 
115 
2-78 
0-99997 
12-0 
132 
150 
2*22 
0-99996 
15-0 
192 
209 
l *67 
0-99996 
18-0 
265 
277 
l-ll 
0-99997 
21-0 
343 
353 
0-56 
0-99998 
24-0 
430 
437 
0-00 
1-00000 
1-00000 
27-0 
526 
527 
30-0 
626 
623 
A glance at the two columns shows that the volumes do not agree to within the limits 
of error above indicated, the numbers being approximately the same for the higher and 
lower but widely divergent for intermediate temperatures. Mr. Hubbard’s observations 
were made by a method substantially the same as our own, but the results are affected 
by a circumstance which might at first sight be supposed to conduce to their accuracy. 
Mr. Hubbard employed unusually large volumes of sea-water in his experiments ; but it 
should be noted, as indeed Kopp has already pointed out, that in such observations the 
increased difficulty of obtaining a uniform temperature throughout the entire mass more 
than compensates for the accuracy arising from the greater displacement of the liquid 
column. The stems of Mr. Hubbard’s instruments were not graduated, the readings 
being taken from attached scales, probably of wood or metal. Moreover, in the reduc- 
tion, the inaccurate coefficient of expansion of mercury given by Dulong and Petit was 
employed. These circumstances may probably serve to account for the discrepancies 
between the two sets of observations. At all events we may point out that it would 
seem almost impossible that errors of the magnitude of the discrepancies we are dis- 
cussing can occur in our own results. Taking as an example the particular case of 
15° C., where the difference between the two values is large, and referring to Table XI., 
