78 Proceedings of the Royal Society 
results are clearly to be relied on implicitly. In the experiments 
with the sea bucket, also, excellent results have been obtained. 
The results obtained by both methods of experimenting will be the 
more accurate the more uniform the temperature of the water. The 
temperature, especially of the bottom water, has also frequently 
been determined by bringing up a quantity of the mud, and taking 
its temperature when it arrives on board. This method also gives 
very satisfactory results when a considerable quantity of mud is at 
disposal. 
Self-Registering Thermometers . — By far the greatest number of 
observations has been made with self-registering thermometers of 
one form or another. 
The first self -registering thermometer was made by Cavendish.* 
He constructed both a maximum and a minimum thermometer, and 
they were of the kind called by the French a deversement , out- 
flow thermometers. In fact, his maximum thermometer is in every 
particular identical with that known in France as Walferdin’s; 
his minimum is on the same principle, but has a U-formed stem 
instead of a straight one. The disadvantages of this form of ther- 
mometer are two — namely, the indications are not continuous, but 
by jerks, depending on the size of the mercury drops, and they 
require to be constantly set, the maximum at a higher and the 
minimum at a lower temperature than the one to be observed ; they 
also require constant comparison with a standard. They are, there- 
fore, not suitable for use where many observations have to be made 
expeditiously. 
In the year 1782 Sixf published a description of the combined 
maximum and minimum thermometer which bears his name, and 
which has since continued to assert its place among meteorological 
instruments as perhaps the best self-registering thermometer. The 
instrument is too well known to require particular description. It 
may, however, be noted that Six himself did not use a hair for a 
spring to keep his indices from falling down, but a fine glass thread 
soldered to the top of the index, and sticking up in a direction very 
slightly inclined to that of the length of the index, so that it pressed 
gently against the sides of the tube. The advantage of the glass 
* Phil. Trans., 1758, 1. p. 308. 
f Phil. Trans., lxxii. p. 72. 
