— 234 — 
Looking at the D-curve for May 18 we see that the 
westerly maximum is reached one hour later, than in the 
normal curve, the easterly maximum being two hours in 
advance. 
Such deviations from the normal curve are not uncommon 
and we are not allowed to take the differences between the 
values for May 18 and the normal and to consider these 
residuals as being connected with the eclipse. But the 
deviation from the normal course exhibited by the curve for 
the part of it which extends between the two turning points 
just mentioned we may trust to be the effect of the eclipse. 
his deviation is very conspicuous being an accelerated 
easterly movement after first contact, followed by a slackening, 
resulting in a slightly westerly movement, before the middle 
of the eclipse. 
It seems to be most in accordance with the course of the 
curve to accept the effect to be a westerly deviation increasing 
and decreasing with the phase of the eclipse. 
If we assume an easterly deviation in the first part and 
a westerly one in the second part of the eclipse, we do not 
meet with a lagging behind the phase of the eclipse, but on 
the contrary a going in advance, which may be considered 
to be less probable. 
The H-eurve is like the Karang-Sago one, as marked above, 
but the effect of the eclipse is less developed. 
Just as for Karang-Sago we may easily draw a curve 
representing the undisturbed course (see dotted part of the 
eurve in the plate) and measure the departure of the true 
curve from it. 
We obtain an increase amounting to 3 y before the middle 
and a decrease 5 Heer E 
of the eclipse; the whole range amounting thus to 6 7. 
The deviation of the Z-curve from the normal (see also 
curve in plate first report) reaches an amount of 4 y three 
quarters of an hour after the middle of the eclipse. 
