334 JOURNAL OF THE WASHINGTON ACADEMY OF SCIENCES VOL. 13, NO. 14 
(2) Wind directions in the free-air estimated upon the basis of the gradient 
wind relations, were compared with wind directions actually observed by 
means of pilot balloons, and it was found that the estimated wind was correct 
in 63% of the cases and the direction-error did not exceed 90° in 89% of 
the cases. It was discovered further that the average wind velocity for the 
various classes of error decreased markedly with increase of error, thus in- 
dicating that large errors of estimation were associated with very light winds 
and were, therefore, of little practical significance. 
One of the outstanding uses of these charts is in connection with the 
supplying of free-air information to aviators at times when, through inter- 
ference by large areas of clouds, pilot balloon observations are not possible. 
The practical utility of the charts for forecasting general weather must be 
determined after subjecting them to continuous comparison and study by 
experienced forecasters. The experience of the Japanese, however, with 
similar maps constructed somewhat differently has indicated that there are 
advantages for general forecasting. The fact that the degree of reliability 
of these maps is high, as indicated by the study of their accuracy, shows that, 
in any event, they afford a basis for studies relating to the physical processes 
within the lowest 2 kilometers of the atmosphere. 
The second paper, Air Navigation, was presented by Mr. J. P. Aur. 
The paper was illustrated with lantern slides, and was discussed by Messrs. 
Hey, Lirrunenaues, and HumMPpHREYs. 
Author’s abstract: The method which has been used for the most part up 
to the present time in air navigation has been by dead reckoning. For 
cross-country flying and short flights over the water, a good compass and a 
good map are the two most important instruments required. 
For long-distance flying, however, the aviator should be able to locate 
his position by some other means as, for example, by astronomical observa- 
tions or by directional radio bearings, when he is unable to see objects on the 
ground during night flights or while flying above clouds or fog, or ocean. 
The present paper described work which was done during the latter months 
of 1918 at Langley Field in the attempt to develop methods and instruments 
for navigating airplanes by astronomical observations. The problem is 
essentially the same as at sea to determine the position of a ship. Various 
experiments were made using the natural horizon, either land, sea, cloud, 
or haze, but these horizons will not always be available for the aviator. 
Some form of artificial horizon must be provided. Observations with a 
preliminary pendulum-type horizon gave results where the probable error 
of a single observation of the altitude of some celestial body amounted to 
+12’ on the average. 
Various methods of computing the results were tried, and special instru- 
ments for plotting the resulting position-line were developed. 
A sextant with some form of artificial-level attachment, either pendulum 
or bubble, was found to be superior to the preliminary pendulum-type 
artificial horizon. 
It became apparent that, when only one celestial body is available for 
observation, it would be necessary also to measure the azimuth of this body 
in addition to the altitude. Experiments along this line gave surprisingly 
good results. Azimuths of the sun could be determined with a probable 
error of +0.3 for means of 10 observations. Thus with some of the methods 
of reduction which are available, it will not be necessary for the aviator to 
know his dead-reckoned position very accurately jf he is able to measure 
both the azimuth and the altitude of the celestial body. 
