the evidence furnished by the cumulus clouds that form during the day between 1000 and 2000. metres, and disappear at night, thus visibly indicating an increase of humidity by day and a decrease by night. If the trend of the humiditycurve at a height of 1000 metres is assumed to be the reverse of its trend at the ground, then the results from the kite-meteorograph show the minimum humidity to be at the coldest and the maximum humidity at the warmest part of the day. The mean daily ranges for different altitudes are plotted in Plate IX., Fig. 2. The part of the curve at the left of the zero line shows the range at different altitudes, with the minimum humidity near the warmest time of day, while the part at the right of the zero shows the ranges at different altitudes, with the minimum humidity at the coldest time of day.

Types of Change of Temperature with Altitude.When the records of temperature and humidity made aloft by the kite-meteorograph and at the stations near the ground are plotted in relation to altitude, they are found to be easily divisible into a few types. In Plate X., Type I represents the decrease of temperature on most fair days from the ground to altitudes of a mile or more, when no clouds are met. The continuous line, plotted from the records of the ascent, represents the day

PLATE X.-CHANGES WITH HEIGHT RECORDED BY KITES AT BLUE HILL.

conditions, and the broken line, plotted from the records of the descent, represents the night conditions. This curve shows that with increasing altitude the temperature falls uniformly during the day and approximately at the adiabatic rate represented by the dotted lines. The fall of temperature with increasing altitude during the night is slower than during the day, and in fact, from the earth's surface to an altitude of a few hundred metres, there is often a rise of temperature with height, so that the air at altitudes of from 300 to 500 metres may be considerably warmer than it is at the ground. This was shown in the descent on October 8, 1896, and is found in Type 3.

When clouds are traversed during the flight, the temperature curve assumes the form of Type 2. The continuous curve is plotted from the records of an ascent; the broken curve from the records of the descent, both occurring in the day-time. The temperature falls at the adiabatic rate in unsaturated air till the base of the cumulus cloud is reached. It falls at a slower rate in the cloud, the rate probably being that computed by physicists as the adiabatic rate for air in which condensation is taking place. Above the clouds, the fall of temperature appears to be very slow.

Type 3 is a condition which persists throughout the day and night, and it resembles the night form

of Type I. The temperature rises very rapidly for a short distance above the ground and then falls, with increase of height, somewhat slower than the adiabatic rate. The rise of temperature near the ground with increasing height is more marked after sunset than during the day-time.

Type 4 was illustrated by the ascent of October 8. This distribution of temperature is caused by a warmer current overflowing colder air, which is very commonly found at low altitudes in the atmosphere and probably exists usually at some altitude, great or small. Recent observations indicate that this type represents the normal condition of the atmosphere in all sorts of weather. Frequently there are two or more sudden rises of temperature at different heights, so that the plotted data resemble inverted stair-steps. During the day there is a decrease of temperature at the adiabatic rate (1°8 in 100 metres) from the ground to the height of several hundred metres, then a sudden rise of temperature in the next one or two hundred metres, and above this a slow fall of temperature with increasing altitude, usually much less than the adiabatic rate. Generally, clouds are found near the plane of meeting of the warm and cold current.

The reverse of Type 4, that is, a sudden fall of temperature, due to a colder current overlying a warmer one, is probably impossible, because the

colder air, on account of its greater weight, would immediately begin to sink and the warmer air would rise. This should cause a fall of temperature at the adiabatic rate from the ground to the top of the colder current, and is probably the origin of the "cold wave" shown in Type 5. Both the continuous and broken curves (representing an ascent and a descent) show a fall of temperature at the adiabatic rate of unsaturated air, from about 500 metres to the highest point reached. Up to 500 metres the decrease of temperature is more rapid than the adiabatic rate, due to the rapid moving in of colder air above, whereby air rising from the ground is cooled by contact as well as by its expansion, and also because the air is heated more than usual by contact with the ground, which under these conditions is abnormally warmer. This is the special characteristic of the "cold wave" type of curve during the day hours. The night form of Type 5, notwithstanding the excessive radiation from the ground through the dry air, shows a rapid decrease of temperature with increase of altitude from the ground upward.

Type 6 shows a less common, but an interesting form, of vertical distribution of temperature, in which the temperature is about the same from 400 metres to 1400 metres or more. Up to 400 metres there is a fall of temperature with increasing alti