81 
of Edinburgh, Session 1878 - 79 . 
Negretti & Zambia, satisfies the conditions required of a thermo- 
meter for isolated observations as completely as can be hoped for. 
It is a mercurial thermometer ; the bore of the stem is contracted to 
the smallest possible diameter at a point about an inch from the 
neck of the bulb. As long as the thermometer is standing vertically 
stem uppermost, the mercury is continuous in stem and bulb, but if it 
be inverted the mercury parts at the contraction, the portion in the 
stem falling down into the point. The stem is graduated from the 
point towards the bulb, and the temperature at the moment of 
inversion is read off by the height of the mercury in the end of the 
stem. This thermometer exists in two varieties, the one with a 
straight stem, which registers by simple inversion, the other with a 
U-formed stem, which requires to be turned completely round. 
The turning arrangement for the latter instrument is a somewhat 
elaborate and expensive instrument, but it answers its purpose. 
The inverting arrangement, supplied with the half-turn thermometer, 
is somewhat clumsy and unsatisfactory. The half-turn instrument, 
when fitted with a suitable inverting arrangement, is to be preferred 
to the others in all work at moderate depths. For ocean work it 
would probably be necessary to give up the protection of the whole 
stem, as it would be impossible to guarantee a tube, which can 
contain the whole instrument, against collapse when exposed to 
pressures of over 500 or 1000 fathoms of water. If the bulb and 
the twist on the stem were protected it would be quite sufficient. 
Sources of Error in the Indications of various Thermometers . — 
When an ordinary thermometer, protected by badly-conducting 
envelopes, is used, it is obviously exposed to alteration of tempera- 
ture by being pulled through warmer or colder water on its way to 
the surface. Whether any sensible error is likely to result from this 
cause must be determined in each particular case by experience. 
The more perfectly it resists change of temperature the longer it 
will take to assume the temperature of the water. Saussure left his 
thermometer down for twelve or fourteen hours for each observa- 
tion, so that this method is now seldom used. Similarly, also, 
the method which depends on bringing up a sample of the water in 
a vessel fitted for the purpose, and taking its temperature with an 
ordinary thermometer when it reaches the surface, has been discon- 
tinued, for although it does not take much more time than would 
