[A] When nitro-sulphuric acid is used, the spark is more powerful, but
  local chemical action can then commence, and proceed without requiring
  metallic contact.

957.  Being thus arranged, there was no chemical action whilst the plates were not connected.  On making the connexion a spark was obtained[A], and the solution was immediately decomposed.  On breaking it, the usual spark was obtained, and the decomposition ceased.  In this case it is evident that the first spark must have occurred before metallic contact was made, for it passed through an interval of air; and also that it must have tended to pass before the electrolytic action began; for the latter could not take place until the current passed, and the current could not pass before the spark appeared.  Hence I think there is sufficient proof, that as it is the zinc and water which by their mutual action produce the electricity of this apparatus, so these, by their first contact with each other, were placed in a state of powerful tension (951.), which, though it could not produce the actual decomposition of the water, was able to make a spark of electricity pass between the zinc and a fit discharger as soon as the interval was rendered sufficiently small.  The experiment demonstrates the direct production of the electric spark from pure chemical forces.

[A] It has been universally supposed that no spark is produced on making the contact between a single pair of plates.  I was led to expect one from the considerations already advanced in this paper.  The wire of communication should be short; for with a long wire, circumstances strongly affecting the spark are introduced.

958.  There are a few circumstances connected with the production of this spark by a single pair of plates, which should be known, to ensure success to the experiment[B].  When the amalgamated surfaces of contact are quite clean and dry, the spark, on making contact, is quite as brilliant as on breaking it, if not even more so.  When a film of oxide or dirt was present at either mercurial surface, then the first spark was often feeble, and often failed, the breaking spark, however, continuing very constant and bright.  When a little water was put over the mercury, the spark was greatly diminished in brilliancy, but very regular both on making and breaking contact.  When the contact was made between clean platina, the spark was also very small, but regular both ways.  The true electric spark is, in fact, very small, and when surfaces of mercury are used, it is the combustion of the metal which produces the greater part of the light.  The circumstances connected with the burning of the mercury are most favourable on breaking contact; for the act of separation exposes clean surfaces of metal, whereas, on making contact, a thin film of oxide, or soiling matter, often interferes.  Hence the origin of the general opinion that it is only when the contact is broken that the spark passes.