ME. J. PEESTWICH ON SUBMAEINE TEMPEEATUEES. 
603 
and a series of twenty-three interesting experiments were made by MM. Martins* and 
Bra vais between the North Cape and Spitzbergen, and off the west coast of that island, 
in depths of from 200 to 2460 feet. 
The principle of overflow differential thermometers had been revived by WALFERDiNf 
in 1836 — a maximum one for the purpose of taking the higher temperatures of deep 
wells and mines, and a minimum one for deep-sea soundings. These instruments were free 
from the inconveniences of Cavendish’s, were of easy manipulation, and could bear jerks 
without affecting the registering column of mercury. To protect them against pressure 
they were enclosed in a tube of glass, of thickness proportional to the pressure to which 
it would be exposed, and hermetically sealed at both ends. M. Walferdin claimed 
for these thermometers greater accuracy and certainty than the ordinary self-registering 
thermometers J. 
These thermometers, termed “thermometres a deversement,” were used by Martins 
and Bravais on their voyage to Spitzbergen, in conjunction with Six’s thermometers 
(thermometrographes) modified by Bunten, of Paris. The former were enclosed in 
glass tubes exhausted as much as possible, and the latter in copper tubes, evidently 
not strong enough, as they “almost always came up full of water.” To ensure accuracy, 
they employed in all these observations two instruments of each sort, and in some cases 
as many as four, and took the mean of each set. When sunk to the bottom they were 
raised 1 metre from it, and left there for an hour. Sometimes the thermometrographs 
were not protected ; and in that case, or when the tubes were full of water, a correction 
was applied, of which we shall speak further on. A correction was also used for the 
angle of the rope with the vertical. M. Martins states that he had much more confi- 
dence in Walferdin’s thermometers than in Bunten’s. I find, however, that, taking 
the 18 observations made with sets of the former, the average variation for each set 
amounted to 0 O- 45 Cent., or, averaging the variation of each of the 52 instruments 
employed, to 0°T6 C., while the 10 observations with 23 instruments of the latter give 
respectively 0°T8 C. and 0 o, 08 C. ; but M. Martins shows that while the mean of the 
differences is 0°T9 C. at depths not exceeding 131 metres, it is reduced to 0 o, 06 C. at 
depths of 640 to 870 metres. The readings, on the whole, of Walferdin’s instru- 
ments are very slightly lower than those of Bunten’s ; as they were more relied on by 
the observers, I have given them in the Tables in preference to the others. 
But notwithstanding the successful use of Walferdin’s instruments on this voyage, 
* Voyage de £ La Eecherche,’ Geogr. et Phys. vol. ii. (Memoire sur les Temperatures de la Mer Glaciate a la 
surface, a des graudes profondeurs, et dans le voisinage des glaciers du Spitzberg, par M. Charles Martens) 
pp. 342-5. Tableau IV. f Bull. Soc. Geol. de France for 1836, vol. vii. pp. 193 & 354. 
t He instances a case of a well at Saint-Andre where, at a depth of 830 feet, two of his instruments gave 
17 0, 96 C. and 17 0, 93 C. respectively; whereas two self-registering instruments gave 19°-2 C. and 16° - 8. The 
latter were affected both from water getting into the case and from lowering of the index by shaking. In 
another case, two of his instruments both registered 23 0, 5 C., and two thermometrographs 23°-45 and 23 o- 50, 
while another of the latter had its index displaced by the shaking of the line. — Ibid. vol. ix. p. 255, vol. xii, 
p. 166, and vol. xiii. p. 113. 
MDCCCLXXV. 4 M 
