It is apparent that our observational knowledge of the atmosphere is gained by two general methods of exploring it, viz. observations made from the earth upon clouds and optical phenomena at a distance, and observations made directly in the air itself. Although it was realized at the beginning of this century that meteorological observations were almost all conducted at the very bottom of our atmosphere--"in the shoals and shallows of the ocean of air," von Humboldt said--yet only within the past thirty years was it thought necessary to replace the occasional observations on mountains by systematic and long-continued ones, comparable to those so generally carried on at low levels. It is an evidence of the zeal in America to advance the young science of meteorology, that the first mountain-top station in the world was established in 1871 upon Mount Washington, and that both this exposed post of observation, 6300 feet above the sea, and the one more than twice as high on Pike's Peak, which was for a long time the highest in the world, were maintained for many years by the United States Signal Service. The present highest station in the world is maintained by the Harvard Observatory upon El Misti in Peru, where, at a height exceeding 19,000 feet, a combination of self-recording instruments was constructed by my assistant, Mr. Fergusson, to operate during three months without attention. It must be admitted, however, that the addition to our knowledge of the physics of the atmosphere afforded by the American stations has been slight and incommensurate with the expense incurred. More has been gained from the mountain stations in Europe, notably from those in the Austrian Alps, which have furnished data for Dr. Hann's splendid discussions of the thermo-dynamics of the atmosphere. While mountain stations present the only means of obtaining continuous observations at a considerable and constant height, still they have serious drawbacks. Not only is the distribution of mountains over our globe irregular, but since they form part of the earth's crust, terrestrial influences affect all observations made upon them. In the case of plateaux this was at once admitted, but by placing the stations on the summits of high and isolated peaks, it was hoped to approximate to the conditions of the free air. It is now recognized that the equilibrium of the atmosphere is so delicate that for its dynamical study exact and minute measurements of temperature, moisture, and currents are required, and the methods which will be described are intended to give the values of these elements free from terrestrial disturbances.