608 TRANSACTIONS OF SECTION G. 
most accurate observations have been obtained with signals sent out by the 
observatory at the Hiffel Tower at 10.45 a.m. and 11.45 p.m. ‘The method adopted 
in the earlier tests was to use a ‘ Perikon’ detector in series with galvanometer 
and telephones, the measurement of strength being made by the cumulative 
deflection due to a series of known signals. ‘This method was not found satis- 
factory with the Paris signals for which the antenna current used was known, 
and in the later tests an Einthoven string galvanometer has been employed by 
which the strength of signal for each individual spark at Paris could be observed 
to within + 5 per cent. The results obtained show that there is a maximum 
variation from 0°6 to 1°3 in the strength of the signals received on different days 
in the same month; the average strength of signal being assumed to be 11, and 
that the current received on a fine, clear night is about 1°7 times as strong as 
that received in the day-time. 
Although no certain relationship can yet be regarded as established between 
the strength of a signal and the weather conditions at the sending and receiving 
stations, so far, observation has shown that rain in Paris always corresponds 
with a diminution in strength of received signal. In one case, with a wind of 
six metres per second velocity, blowing in a N.W. direction, the signal-strength 
fell to half its normal value. The most favourable condition for signalling 
appears to be a cloudy sky at both sending and receiving stations, the signals 
being weaker when the sky is clear or covered with light clouds. Rain at the 
receiving station appears to have a comparatively small influence on the strength 
of the received signals. 
The result of a set of special signals sent from the Eiffel Tower on the even- 
ing of Saturday, July 26, 1913 (by the courteous arrangement of Comm. Ferric), 
at intervals of 30 minutes, between 7 and 10 p.m. (which includes the time of 
sunset), shows that the increase in strength of night signals occurs about three- 
quarters of an hour after sunset, there being a sudden increase in strength of 
about 70 per cent. This change is quite sudden, there being comparatively little 
alteration in signal-strength until darkness has set in, and no_ perceptible 
increase in strength afterwards. There appears to be some evidence that signals 
are slightly stronger just after sunset than during normal night conditions. 
6. Short Heat Tests of Klectrical Machines. 
By W. BR. Cooprr, M.A., B.Sc. 
Tests of dynamo-electric machinery are generally carried out extending over 
six hours in order to determine the maximum temperature rise. Suggestions 
have been made that such tests might be considerably shortened by assuming 
that the curve of temperature rise is a logarithmic curve. The author gives a 
brief account of the methods that have been proposed, and points out that the 
‘thermal time constant,’ on which the behaviour of a body in heating and cooling 
largely depends, should be found most easily, and under less complex conditions, 
from the cooling curve. In order to test the applicability of the various 
methods, tests were made upon a 5 kw. motor-driven dynamo. The curves of 
temperature rise were found to be fairly logarithmic. The usual method of 
running a machine on test is to run on constant output with constant voltage, 
which necessitates increasing the input to the field coils as their resistance 
vises. A truly logarithmic curve can only be expected if the input of heat 
is at a constant rate, so that a better result would be expected if a machine 
were tested with constant input to the field coils, the output of the machine 
being constant but at varying voltage. Actually this method of testing was 
not found to give an improved result. Graphical methods appear to give 
better results than formule, as the latter are very sensitive to small errors in 
the data. In any case only certain portions of the temperature-rise curve should 
be used, and any formule depending on the initial rate of temperature-rise 
should be avoided, as this is difficult of exact determination. The curve of 
cooling has certain advantages; but only a small portion of it seems suitable for 
graphically determining the thermal time constant, and this quantity can be 
derived much more readily from the time taken for a certain percentage drop of 
temperature-rise. In the results given there is better agreement between the 
values found for the thermal time constant than for the maximum temperature- 
