955 
balloons during the war, a difficulty which was solved only 
later. The observations with kites and cable balloons at Soester- 
berg (S) continue to suffer strongly under the absence of a private 
ground, the negotiations for which only recently met with success. 
By way of supplement, therefore, under the heading “Distribution of 
wind and temperature’, we have resorted to observations at Hamburg, 
Lindenberg and Friedrichshafen, so far as these were at hand at the 
time. The latter have already been published in their definitive 
form’), the former are quoted only partially in the German daily 
weather reports and therefore may afterwards be largely supplemented 
or amended. Afternoon observations are indicated by adding the 
index p to the name of the station. 
On October 8th there was almost no increase of wind from the 
St-Cu level up to the height where the Ci floated, and certainly 
no increasing SW.-wind such as would be required to bend back 
to the earth rays which left the source of sound with some elevation. 
Up to 2000 M. no S.-components are met with, and no nearer than 
Helder a W.-ly wind of some importance. The direction of the Ci, 
observed at Flushing, is nearly perpendicular to the principal direc- 
tion of propagation. Given the position of the anticyclone. in general, 
nothing else was to be expected. Hence the abnormal audibility as 
far as 220 KM. cannot be ascribed to an increase of wind in the 
troposphere. One might suppose, that in the stratosphere an in- 
creasing SW.-current or a decreasing NE.-current reigned, and that 
these currents caused the rays, practically straight in the troposphere, 
to curve in the stratosphere. The smallest distance, at which the 
rays may return to the earth, is then determined by the height up 
to which one makes the wind increase and by the amount of that 
increase. But in that way we can never get anything else but a 
curvature of the rays, which propagate the sound in the direction of 
the wind. For a ray perpendicular to the wind the effect is practically 
equal to zero, and the limit of the silent region cannot be a circle, 
but must show a smaller, perhaps even an inverted curvature so that 
half a circle, as our figure 9 shows, cannot at all be explained in 
this way (see the footnote on page 941). 
At this moment we indicate once more a remarkable deduction, 
which must be made from the theory of the bending of sound rays 
in consequence of the decrease of temperature with height. In the 
normal ease of fall of temperature with increasing height, the sound 
rays are curved from the surface upwards. Rays starting with a 
1) Ergebnisse der Arbeiten der Prachenstation am Bodensee im Jahre 1914, 
Stuttgart 1915. 
