260 Mr. G. Jatté on the Conductivity of 
then in different parts of the building. In the latter case 
the variations of the natural leak showed no agreement, 
except in a few instances of great changes; but when the 
two electroscopes were kept close together there was an 
exact parallelism between them. The changes occurring 
were not proportional, and as long as they were within the 
errors of experiment (less than 10 per cent.) they did not 
even always show the same direction ; but whenever any 
larger variations took place, a close agreement in the be- 
haviour of the two vessels could be observed. These 
experiments were carried on for about two months; the 
analogy was not always equally good, but on the whole 
unmistakable. Table II. contains the results for ten days of 
observation when the resemblance was particularly good. 
The amount of ionization is given in arbitrary units, but 
chosen in such a way that the averages for the two cases 
coincide. The numbers for the leak in the day are means of 
8 hours, those for the leak in the night means of 16 hours. 
It will be noticed that for this particular period the leak at 
night is always bigger than in the daytime, except. on 
March 12th; but it has alr eady been mentioned that this is 
not always the case, and seems due to special local conditions. 
TABLE II, 
mo i ml 
| Leak in the silvered Leak in the brass | 
Date. glass vessel. | vessel. 
In the day. | In the night. | In the day. | In the night. 
ee mal —S ae et 
March 10 ...... LP 67 1 aoe 7-0 
wh DS ee cc 5:9 6-7 / lip 62 
7 ae 70 | | 79 
¢ dedi, 6:0 68 | 58 66 
>. «lee 60. |), 64 4 hs aa 
= ee 60. 6:9 | 52 6:8 
: 1 (aaa 6:0 6:8 56 6°5 
be 18 we. ess awe 6'6 54 | 6°5 
ec) me 56 ~ 5:0 | 
ss 4, ee 6:3 rit 6:2 Ts 
ae 62 - 6:2 | 
The parallelism between the two vessels is not limited to 
the average values for day and night, but was also observed 
for the variations occurring during the day, though to a less 
extent, as might be expected from the less accuracy of read- 
ings over a short range of time. There was also a distinct 
indication of a slight periodical diurnal variation with the 
minimum between noon and 2 P.M., but this could not always 
