upper part of the diagram, is slower at night than during the day. It appears that the diurnal change of temperature is very small at great altitudes, compared with the change near the earth's surface. The relative humidity (Fig. 5) up to 2000 metres varies inversely with the temperature, and in the present case there was only a slight change in the direction of the wind (Fig. 6). [Illustration: PLATE IX.--MEAN CHANGES WITH HEIGHT, AND CHANGES DURING THE KITE-FLIGHT OF OCT. 8, 1896.] =Diurnal Changes of Temperature at Different Altitudes.=--The curve representing the diurnal change in the air at some distance above the ground is probably similar to one representing the change near the ground, except that its amplitude is less. If this be true, then the diurnal rate of fall for a given time at any two levels will be proportional to the daily ranges of temperature at the two levels. It is impossible in practice to keep a kite at exactly the same level for twenty-four hours; hence the daily ranges for the different levels must be found by comparing the rates of rise or fall of temperature for given times with the rates found from records near the ground, made simultaneously with those above. In Plate IX., Fig. 1, the results for six stations, _i.e._ the kite at 1000 and 500 metres, the Eiffel Tower in Paris (300 metres), the summit of Blue Hill, its base, and the valley (200, 50, and 15 metres respectively), are connected, and a smooth curve is drawn through them. The curve passes approximately through every one of the observed and the computed ranges, except the one at the summit of Blue Hill, which is too great. This evidently is because insolation and radiation, acting through the soil of the hill, heat and cool the air to a greater extent than the free air is heated and cooled at the same altitude, and this must be true at every mountain station. The smoothed curve passes also very slightly to the left of the data for the Eiffel Tower, indicating that the range there is about 1∞ greater than the true range on account of the heating and cooling of the Tower. From this it appears that the diurnal range of temperature diminishes rapidly with increasing altitude in the free air, and almost disappears in the average at a height of 1000 metres.