954 
strata, so that they had usually to resort to suppositions or theoretical 
speculations. E.g. DE QUERVAIN in his case assumes, that from 4000 
to 11000 m. height a SW.-wind occurred, increasing to a velocity 
of 35 m. per second, whereas the observation of Cirrus-clouds in 
our country on the 15% gave Ci. in NW. (probably rising from 
NW), and on the 16 and 17' really a rather fast motion from 
WNW. (17 m. at 10000 m.) was stated. Inspection of the weather- 
chart, in connection with the motion of the isallobars, would lead 
one in the first place to sappose a motion from SE. in the inter- 
mediate layers. If the sound rays have really taken their way ex- 
clusively through the lower atmosphere, we should be more inclined 
to admit the influence of a streng inversion in the A-Cu-level in the 
same manner as was obseryed on that day at a much lower level at 
Lindenberg. Be this as it may, observations of the air-motion in the 
levels above 4000 m. over Switzerland fail on this occasion. 
Fortunately, for some of the cases of roar of cannons in the Nether- 
lands treated above, a greater certainty exists. We have collected in 
a table the observations obtained from clouds, and by means of 
pilot-balloons, kites and cable-balloons for the days mentioned, and shall 
consider how far the particularities of the sound phenomena may 
be explained by these observations. 
The table gives in the first column the dates, in the second a 
short review of the weathereonditions, principally ‘he distribution 
of pressure. Under the heading “Wind- and cloudmotion” then follow 
the observations of wind and clouds at the stations Helder (He), 
Flushing (VI) and Winterswijk (Wi), where the observations of clouds 
have been made since 1897 and have been made daily since 1905, whereas 
at De Bilt (De B) they are made on special occasions. The direction 
and relative velocity are determined accurately by Brsson’s nepho- 
scope, the height of the clouds has to be estimated according to the 
type of cloud and has been put for simplicity equal to 1000 m. 
for the lower clouds, 2000 m. for the Strato-Cumulus (St. Cu), 8000 m. 
for the Alto-Cumulus (A-Cu) and 10000 m. for the Cirrus (Ci). By 
these assumptions we are enabled to calculate the velocity in meters 
per second. In consequence of variations in height for the lower 
clouds, the real velocities may differ from the calculated ones by the 
factor 0.4 to 1.5; for the higher clouds the deviations will probably 
remain below 20 °/,. 
Evidently it was desirable to supplement these estimates by more 
exact measurements, which might consist in observations with pilot 
balloons, cable balloons, or kites. Unhappily enough the first mentioned 
observations have suffered much by the difficulty of obtaining pilot 
