cerning the movement of the atmosphere in that region. It is apparent at the outset that the lower limit of the upper inversion is not sharply defined, but that the air motion in the explored part of that region, at least, partakes of, and probably is controlled by, that of the lower levels of the atmosphere on which it rests. The observations of the wind velocity in the region of the upper inversion were not conclusive in any respect, other than that the movement was at times considerable, and again of rather low value, as on September 7, 1910, when, at Huron, S. Dak., winds at 3 to 11 kilometers altitude (1.9 to 6.9 miles) averaged about 16 meters per second (36 miles per hour). At the base of the upper inversion a wind of 18 meters per second (40 miles per hour) was encountered; at 1,000 meters (3,280 feet) higher, the wind had increased to 32.5 meters per second (73 miles per hour). It continued at a high velocity up to 17,227 meters (10.7 miles) and then suddenly fell off to 6.8 meters per second. (15 miles per hour). On another occasion, September 4, 1910, the enormous velocity of 42.2 meters per second (95 miles per hour) was found at the base of the upper inversion, and a still higher velocity of 48.5 meters per second (108 miles per hour) was encountered somewhat higher. Above this, however, the speed of the wind diminished to zero. The ascension of the 4th was in a cyclonic area, while that of the 7th was on the front of a strong anticyclone moving toward Huron from the British northwest.

Another interesting conclusion that may be drawn from the sounding-balloon ascensions, and also from observations on high mountain stations, is that the gyratory motion of the air characteristic of cyclones at the surface and for some distance above, does not extend far upward. The movement of the upper layers, say above 10,000 meters (about 6 miles), as indicated from the drift of balloons that ascended to that altitude, appears to be in three main directions, viz, from west to east under normal conditions; from north to south, or northwest to southeast, when anticyclones dominate the weather; and from south to north, or southwest to northeast, when cyclones control the weather. Perhaps a better way of expressing the idea would be to say that the air currents are from some northerly direction on the east side of anticyclones and from some southerly direction on the west side, and that under practically all other conditions the drift of the air in the very high levels is from west to east.

One of the interesting facts brought out in connection with ascensions in anticyclonic conditions is that the prevailing west winds of the middle latitudes, formerly believed to extend in an unbroken stratum from an altitude of about 5 kilometers (3.1 miles) to at least 16 kilometers (10 miles), are at times wholly suspended up to an altitude of 12 kilometers (7.5 miles). This fact is confirmed by observations made on Pike's Peak, Colo., as will be referred to later.

In 39 ascensions made under the direction of Prof. Rotch, in which the altitude reached was 6 miles or over, 11 balloons landed almost due east of their starting point, 22 landed south-southeast of their starting point, and 6 landed north-northeast of their starting point. It is not always, nor in the majority of cases, possible to tell from surface conditions the direction the balloon will take. Sometimes,

1 Bigelow reached the same conclusion from a study of cloud observations. See Report Chief of Weather Bureau, 1898-1899, p. 434.

however, there is fair agreement between surface pressures and upper wind drift. In general there is a northerly component in the winds in front and on the east side of an anticyclone, although numerous exceptions to this rule have been noted. One of the most marked exceptions was on November 25, 1904, when a balloon launched at St. Louis, Mo., traveled almost due east to near Louisville, Ky., although the pressure distribution at the surface clearly indicated northerly winds, and winds from that direction actually prevailed at the ground. This balloon, which reached an altitude of 11,500 meters (7.1 miles), and the one sent up the following day, moved with the enormous average velocity of 100 miles per hour. The second balloon, instead of moving toward the east, as did the one launched on the previous day, moved in a south-southeast direction and landed in western Tennessee. From this change in direction of the air currents it is evident that some temporary disturbance occurred in the atmosphere sufficient to modify greatly the eastward flow. What the disturbance was is not apparent from surface conditions. On the day that the balloon moved eastward there was a marked barometric depression over southern New England which had been stationary for about 24 hours. It may have been that the pressure in the higher levels over New England was falling on the day in question, and that the high eastward velocities encountered by the balloon were due to a pressure gradient that existed in the upper regions only. Hann showed more than 20 years ago that atmospheric pressure on mountain tops continues to fall for some time after the turn to rising pressure has set in over surrounding low levels.

The cause of the changes in the direction of the wind aloft is not always apparent from surface distribution of temperature and pressure. Primarily, the direction of the wind on the earth's surface is dependent on the temperature and pressure, the winds blowing from regions of low to regions of high temperature and from regions of high to regions of low pressure. In the United States the strong winds of winter have regions of higher temperature on their right and in slightly higher latitudes. Unfortunately we are not able to study the temperature changes in the atmosphere as a whole, but only in a thin stratum next to the earth's surface. The upper winds in the United States are uniformly from the west, as has been fully demonstrated in the past. That these prevailing westerly winds are subject to important modifications is shown by the motion of the upper clouds and by the travel of sounding balloons.

Sounding-balloon ascensions have added very much to our knowledge of the temperature of the atmosphere up to heights of 15,000 meters (9.3 miles) and even higher, but the number of ascensions to heights above 9.3 miles is as yet small.

The vertical distribution of temperature in different sections of the same anticyclone is well shown by the simultaneous ascensions at Omaha, Nebr., and Indianapolis, Ind., on October 5, 1909. The two stations were, roughly speaking, within the influence of a great anticyclone, Indianapolis being nearest the center and under the higher pressure. The pressure at Indianapolis being higher than that at Omaha. we should expect lower surface temperature, as was actually found. But the low temperature of the air column over Indianapolis extended up to 2 km. (1.2 miles) only, at which level the air-column temperatures at the two places were reversed, the western station

becoming the colder at that level, and steadily remaining so up to 14 km. (8.7 miles). During an earlier ascension at the two stations, on September 30, 1909, the surface weather conditions were quite different from those of October 5, the two stations being separated by a shallow anticyclone, with Indianapolis on the eastern edge and Omaha on the western edge. As in the first-named case, the eastern station was the colder up to about 3 km. (1.9 miles), but from that altitude up to about 12 km. (7.5 miles) the Omaha air column was the colder, the difference at the 12 km. level amounting to 16° C. (28.8° F). Marked variations of the temperature at similar great altitudes have been recorded elsewhere, especially in England, where the temperature of the lower limit of the upper inversion has been found to differ on the same day as much as 20° C. (36° F.) at stations not more than 150 miles apart. The lowest temperature recorded in any of the Weather Bureau series of ascensions was 68.9° C. (-92° F.) at Huron, S. Dak., in September, 1910.

A study of observations at mountain stations in Colorado has shown that variations of temperature at the summit and at the base stations are nearly coincident in point of time and are generally similarly directed, but that there are occasions when a fall in temperature sets in on the plains while the temperature on the mountain tops is still rising. In rare cases, also, the weather conditions on the mountain summits are controlled by causes that are not operative on the plains to the eastward. These studies have increased our knowledge of the effect of local topography in the warming and cooling of the air that is trapped between the mountain ranges. The important fact, revealed in connection with sounding-balloon ascensions, that the prevailing eastward drift of the atmosphere is wholly suspended during the prevalence of strong anticyclones is confirmed by a study of the records of wind movement over the high stations of eastern Colorado at Corona and Pikes Peak. There can be no doubt that the local circulation in strong anticyclones up to the level of Pikes Peak is controlled by the anticyclone, though this is seemingly controverted by observations on the movement of high clouds in other parts of the United States.

The cirrus level in the United States is about 15 km. (9.3 miles) above sea level. Clouds in this level have been observed to move directly across the central areas of anticyclones, from west to east, which movement would not be possible did an easterly current prevail at that level. The wind movement over Pikes Peak, Colo., 4,301 meters (14,111 feet) above sea level, is from the northeast when an anticyclone occupies the Great Basin to the westward, thus indicating the local control of the wind circulation by anticyclones at the level of Pikes Peak.

At Mount Weather, Va., the kite flights thus far made show that practically all easterly winds, except under special conditions, are shallow winds; that is, they are generally less than a mile in vertical

extent.

SOLAR RADIATION.

Between July 16 and October 10, 1910, Prof. Kimball was engaged in a pyrheliometric survey of the region west of the Great Lakes and the Mississippi River, preliminary to the establishment of permanent

23165°-AGE 1911- -11

observing stations. One of these, Madison, Wis., has been in operation since July 22, 1910, and others will be equipped as soon as apparatus already ordered is received. Pyrheliometric observations have been maintained throughout the year at Washington, D. C., and were resumed at Mount Weather in May.

The observations at the western stations showed radiation intensities in excess of the five-year averages for Washington, the excess ranging from 4 per cent in August, at Lincoln, Nebr., to 22 per cent in September, at Flagstaff, Ariz.

The most striking features of the year have been the high value of the radiation in February and March on the front of marked high barometric areas, and the low value during the protracted hot wave in May. At Madison, on February 23, and again on March 4, the radiation intensity with the sun shining through an air mass 1.5 (zenith distance of the sun 48°) was 1.67 calories per square centimeter per minute, which is as high as any measurement obtained by the Smithsonian Institution on Mount Wilson during the summers of 1905 and 1906. At Washington the corresponding maximum intensity during this period was 1.47 calories, or 12 per cent less than at Madison.

During the hot wave of May, 1911, the maximum intensity of solar radiation measured at Mount Weather, with the sun at zenith distance of 48°, was 1.20 calories per square centimeter per minute, and the average was little over 1.00 calorie.

Measurements of the polarization of skylight, as well as other considerations, indicate that during protracted hot periods a very considerable percentage of the heat reaching the lower layers of the atmosphere is received diffusely from the sky. A Callendar recording pyrheliometer, capable of measuring the heat thus received, has been in continuous operation at Washington throughout the year; but quantitative results can not be given until this instrument has been carefully compared with a Marvin pyrheliometer, which will be done as soon as a new Callendar instrument provided with an improved form of recorder is received.

In response to the request of certain European investigators, a series of special observations on the positions of the neutral points of Arago and Babinet was made by Prof. Kimball while on field duty. These are now being continued at Mount Weather in connection with the measurements of the percentage of polarization of skylight, made as in previous years.

The five-year averages of solar radiation intensities for Washington were published in the Bulletin of the Mount Weather Observatory, Volume III, part 2. In response to a special request, a résumé of that part of the above paper which treats of sky polarization, together with a summary of the polarization observations obtained while on field duty, was prepared by Prof. Kimball for publication in the Journal of the Franklin Institute for April, 1911.

The constants to equation 20, Bulletin of the Mount Weather Observatory, Volume I, part 4, are being recomputed from data recently furnished by the Smithsonian Institution. Prof. Humphrey's recent computation of the distribution of aqueous vapor in the atmosphere when the sky is cloudless will also be utilized. New tables for facilitating solar constant computations will be prepared from this revised

equation. A copy of these tables has already been requested by the Argentina Meteorological Office.

It is believed that accurate determination of the intensity of direct solar radiation, of the quantity of heat received diffusely from the whole sky, and of the rate at which heat is lost at night will not only be of value to climatologists generally, but will also be utilized by the weather forecaster. Especially urgent is the demand from biologists for accurate data relative to the quantity of heat received from the whole sky. The University of Wisconsin is now furnishing data of this character for use in connection with certain biological studies.

MOUNT WEATHER BULLETIN.

A full discussion of the upper-air observations made at Mount Weather and elsewhere, as well as of the progress made in other special lines of scientific work, will be found in the successive issues of the Bulletin of the Mount Weather Observatory, which, it may be remembered, is devoted to the results obtained from aerial investigations as well as from other special researches into obscure laws of atmospheric phenomena bearing on the physics and mechanics of the whole atmosphere. Although this publication is mostly filled by the results contributed by the staff of the observatory, yet, when space allows it, contributions of fundamental importance presented by other meteorologists are included in the Bulletin.

The completed Volume II, with its index, was issued in July, 1910, and the completed Volume III, with its index, in July, 1911. The second part of Volume IV was sent to the printer in June of the present year.

FORECASTS AND WARNINGS.

The work of forecasting daily weather and temperature changes, storms, cold and warm waves, and frosts-the primary duty of the Weather Bureau-received the careful attention of the corps of forecasters throughout the year. No important meteorological change occurred without notice having been given well in advance.

Storm warnings to Lake, seacoast, and West Indian stations, and frost warnings for the sugar, trucking tobacco, fruit, and cranberry regions, were issued whenever conditions justified. These warnings were successful. Particular attention was given to the hurricanes of September and October, 1910, and a number of testimonials commending the work of the bureau in connection therewith were received. The warnings of the approach of cold waves resulted in a saving of growing crops and prevented injury to many shipments of perishable goods and to farm stock. Daily forecasts of probable wind and weather conditions off the Atlantic coast, eastward to the Grand Banks, were issued for the guidance of transatlantic steamships.

HURRICANES OF THE YEAR.

WEST INDIAN HURRICANE OF SEPTEMBER, 1910.-This storm was first detected near San Juan, P. R., on September 6. It moved in a west-northwest direction, and by the morning of September 14 had reached the Texas coast near the mouth of the Rio Grande. Warnings were issued regularly until the storm disappeared. No loss of

1