In meteorological observations the temperature of the air is of the first importance, and it is a subject to which a great deal of attention is given, thousands of temperature observations being made and recorded every day. It is therefore desirable that these observations shall be as correct as possible. At first sight nothing seems more simple than to take the temperature of the air. All that appears necessary is to hang up a correct thermometer of any construction, anywhere out of the sun, and the thermometer will then indicate the temperature of the air where the instrument is placed. We shall presently see that this is very far from being the case; and not only so, but we shall find that it is difficult to get two thermometers which will give the same readings, when hung near each other in the open air, even though they agree perfectly with each other when placed in water.