Eddington
and Cottingham, those of the second, Crommelin and Davidson. The
conditions were especially favorable, for a very large number of
bright stars were shown on the photographic plate; the observers at
Sobral being particularly lucky in having good weather.

The total eclipse lasted five minutes, during four of which it was
perfectly clear, so that good photographs could be taken. In the
report issued regarding the results the following figures, which are
the average of the measurements made from the seven plates, are given
for the displacements of seven stars:

1''.02, 0''.92, 0''.84, 0''.58, 0''.54, 0''.36, 0''.24, whereas,
according to the theory, the displacements should have amounted to:
0''.88, 0''.80, 0''.75, 0''.40, 0''.52, 0''.33, 0''.20.

If we consider that, according to the theory the displacements must
be in inverse ratio to the distance from the centre of the sun, then
we may deduce from each observed displacement how great the sideways
movement for a star at the edge of the sun should have been. As the
most probable result, therefore, the number 1''.98 was found from
all the observations together. As the last of the displacements given
above--i.e., 0''.24 is about one-eighth of this, we may say that the
influence of the attraction of the sun upon light made itself felt
upon the ray at a distance eight times removed from its centre.

The displacements calculated according to the theory are, just because
of the way in which they are calculated, in inverse proportion to the
distance to the centre. Now that the observed deviations also accord
with the same rule, it follows that they are surely proportionate
with the calculated displacements. The proportion of the first and
the last observed sidewise movements is 4.2, and that of the two most
extreme of the calculated numbers is 4.4.

This result is of importance, because thereby the theory is excluded,
or at least made extremely improbable, that the phenomenon of
refraction is to be ascribed to, a ring of vapor surrounding the
sun for a great distance. Indeed, such a refraction should cause a
deviation in the observed direction, and, in order to produce the
displacement of one of the stars under observation itself a slight
proximity of the vapor ring should be sufficient, but we have every
reason to expect that if it were merely a question of a mass of
gas around the sun the diminishing effect accompanying a removal
from the sun should manifest it