place; either
each end had its own particular spark to the ball, or else one end only was
associated by a spark with the ball, but was at the same time related to
the other end by a spark between the two.

1414. When the ball _c_ was about an inch in diameter, the ends _n_ and
_o_, about half an inch from it, and about 0.4 of an inch from each other,
the two sparks to the ball could be obtained. When for the purpose of
bringing the sparks nearer together, the ends, _n_ and _o_, were brought
closer to each other, then, unless very carefully adjusted, only one end
had a spark with the ball, the other having a spark to it; and the least
variation of position would cause either _n_ or _o_ to be the end which,
giving the direct spark to the ball, was also the one through, or by means
of which, the other discharged its electricity.

1415. On making the ball _c_ smaller, I found that then it was needful to
make the interval between the ends _n_ and _o_ larger in proportion to the
distance between them and the ball _c_. On making _c_ larger, I found I
could diminish the interval, and so bring the two simultaneous separate
sparks closer together, until, at last, the distance between them was not
more at the widest part than 0.6 of their whole length.

1416. Numerous sparks were then passed and carefully observed. They were
very rarely straight, but either curved or bent irregularly. In the average
of cases they were, I think, decidedly convex towards each other; perhaps
two-thirds presented more or less of this effect, the rest bulging more or
less outwards. I was never able, however, to obtain sparks which,
separately leaving the ends of the wires _n_ and _o_, conjoined into one
spark before they reached or communicated with the ball _c_. At present,
therefore, though I think I saw a tendency in the sparks to unite, I cannot
assert it as a fact.

1417. But there is one very interesting effect here, analogous to, and it
may be in part the same with, that I was searching for: I mean the
increased facility of discharge where the spark passes. For instance, in
the cases where one end, as _n_, discharged the electricity of both ends to
the ball _c_, fig. 116, the electricity of the other end _o_, had to pass
through an interval of air 1.5 times as great as that which it might have
taken, by its direct passage between the end and the ball itself. In such
cases, the eye could not distinguish, even by the use of Wheatstone's
means[