— 228 — 
be seen from the behaviour of the H-curve, which at 2% p 
exhibits a secundary maximum, accompanied by an easterly 
excursion of the needle: 
Noon 1h p gh gh 4h 
—0.'1 0.1 0.8 0.0 —0.4 
But this movement towards the East at that hour is 
exactly the particular feature exhibited by the curve of the 
eclipse-day; accordingly we are not allowed to associate it 
with the eclipse without further restrietion. Comparison 
with the eurves of other stations may give more certainty, 
as I already remarked in 1901, but was not able to carry out 
until now, after the appearance of the results of observations 
published by BAUER. 
Consequently I have plotted in the plate the H-curves for 
Colaba, Dehra Dun, Zi-Ka-Wei, Manila, Karang-Sago and 
Batavia after Greenwich-time. 
Looking at these curves we see that the above mentioned 
wave between 16^ and 21^ Greenwich-time is absent in the 
curves for Colaba, Dehra Dun and Zi-Ka-Wei. It is clearly 
seen in the Manila-curve, still more so in the Batavia one 
and most strongly developed in the Karang-Sago-curve, 
accordingly increasing with approach to the line of central 
eclipse. 
Though this is not an absolute warrant for it being an 
eclipse-effect, the probability may however be esteemed to 
be strong enough. 
The D- and H-curves for Karang-Sago of the 13'^ of May 
are quite normal, free from secundary maxima and minima, 
and may be taken as normal ones to compare with the 
eclipse-day curve. 
he D-curve for May 18'^ differs much from this normal 
one. 
As was mentioned in my first report, owing to instrumental 
ill-luck, there is a discontinuity in the D-readings of the 18', 
caused by transition of record on a cylinder rotating in 
24 hours, to a 6-hours cylinder, 
