4 DR HUGH ROBERT MILL ON THE 



except for the slight expansion or contraction of the broken-off column by change of 

 temperature. I made a series of experiments on the temperature correction of eight of 

 these thermometers, and found, as the average of many trials, that a change of 

 60° F. lengthened the broken-off column by 1° of the scale, the temperatures 

 employed being 32° and 92°. Hence for a change of 6° between the temperature 

 of reversing and that of reading, a correction of one-tenth should apparently be 

 applied. In summer, for instance, a thermometer reversed at 45°, hauled up rapidly, 

 and read in air at 65°, might appear to require a correction of o, 3 ; but, recollect- 

 ing that the thermometer is read before it has time to assume the temperature 

 of the air, and that it is wet, we see that the amount of correction should be 

 considerably less. If the thermometer is hung up in the wind for five minutes, and kept 

 wet all the time (for it would be impossible to dry it thoroughly), the wet bulb tempera- 

 ture of the air, ascertained by a sling thermometer, might be used to correct the readings. 

 Or, more simply, the inverted thermometer may be placed for a few minutes, before read- 

 ing it, in a bucket of surface water, the temperature of which can easily be noted, and the 

 corresponding correction applied. This was done on several occasions when the difference 

 of temperature between water and air was over 10°, — a state of matters which rarely occurred 

 in the Clyde Sea Area, except for the bottom temperatures of Loch Fyne and Loch Goil 

 in summer. A simpler method, subsequently adopted, was to read the thermometer as it 

 came up, then right it sharply, and immediately reverse again, the second reading giving 

 the exact temperature of the instrument at which the first reading was made. In prepar- 

 ing the results for publication it was not found necessary to apply this special correction. 

 The thermometers acquired a considerable index error in the time during which they 

 were used : in one case it amounted to as much as 0°'5. This error was determined periodi- 

 cally by the use of melting ice, and of water at a higher temperature in which a standard 

 thermometer was immersed. In calm weather, and by the use of ordinary precautions to 

 insure correct readings, the observations could be trusted to o- l F. Although well suited 

 for work in a climate like that of the West of Scotland, in most of the thermometers I 

 have examined the pouch-shaped recess is not large enough to contain the overflow from 

 the bulb when the temperature is raised 60° or more after inversion ; and in other cases, 

 although the pouch is large enough to hold the overflow, a very slight jerk is sufficient to 

 carry it past the siphon bend, and so vitiate the record. The bulb of the thermometer 

 being filled with mercury and surrounded by it, very rapidly accommodates itself to the 

 temperature of the water in which it is immersed. From experiments made at different 

 times it appears that in ordinary circumstances less than a minute suffices ; but when the 

 difference of temperature exceeds 5° two minutes may be necessary. In order to make 

 perfectly sure of equilibrium of temperature being arrived at between bulb and water, it 

 was usual to leave the thermometer at the depth where it was wanted to register for at 

 least three minutes before causing it to reverse. In the case of the first sounding at any 

 station when the thermometers had remained for some time exposed to atmospheric 

 temperature, an immersion of five minutes was the minimum. 



