sed for the purpose. Moreover, an important
novelty was introduced by the observation of the various colours visible
in the star-couples, the singular and vivid contrasts of which were now
for the first time described.

Double stars were at that time supposed to be a purely optical
phenomenon. Their components, it was thought, while in reality
indefinitely remote from each other, were brought into fortuitous
contiguity by the chance of lying nearly in the same line of sight from
the earth. Yet Bradley had noticed a change of 30 deg., between 1718 and
1759, in the position-angle of the two stars forming Castor, and was
thus within a hair's breadth of the discovery of their physical
connection.[27] While the Rev. John Michell, arguing by the doctrine of
probabilities, wrote as follows in 1767:--"It is highly probable in
particular, and next to a certainty in general, that such double stars
as appear to consist of two or more stars placed very near together, do
really consist of stars placed near together, and under the influence of
some general law."[28] And in 1784:[29] "It is not improbable that a few
years may inform us that some of the great number of double, triple
stars, etc., which have been observed by Mr. Herschel, are systems of
bodies revolving about each other."

This remarkable speculative anticipation had a practical counterpart in
Germany. Father Christian Mayer, a Jesuit astronomer at Mannheim, set
himself, in January 1776, to collect examples of stellar pairs, and
shortly after published the supposed discovery of "satellites" to many
of the principal stars.[30] But his observations were neither exact nor
prolonged enough to lead to useful results in such an inquiry. His
disclosures were derided; his planet-stars treated as results of
hallucination. _On n'a point cru a des choses aussi extraordinaires_,
wrote Lalande[31] within one year of a better-grounded announcement to
the same effect.

Herschel at first shared the general opinion as to the merely optical
connection of double stars. Of this the purpose for which he made his
collection is in itself sufficient evidence, since what may be called
the _differential_ method of parallaxes depends, as we have seen, for
its efficacy upon disparity of distance. It was "much too soon," he
declared in 1782,[32] "to form any theories of small stars revolving
round large ones;" while in the year following,[33] he remarked that the
identical proper motions of the