"Fortunately, Congreve left an account of his activities. This is
contained in two pamphlets (both in the R.A. Library), which he wrote,
entitled "A short account of the improvements in Gunpowder", and "A
statement of facts which have arisen from manufacturing Gunpowder at the Royal Mills since 1783". Taken together, they amount to a record both of what he found and what he did about it. The situation at the time was that the powder makers, who had considerable political influence, had contrived to persuade William Pitt, the Prime Minister, that the Royal Mill at Faversham should be sold to the trade, ending the Board's [Board of Ordnance] involvement in the making of powder, on the grounds that civilian powder makers could make it better and cheaper. On the evidence of the first five years after the Board bought Faversham they were perfectly right. One aspect of Congreve's investigation seems to have been to persuade the government to delay a decision until his
enquiry was completed."
"Congreve's first step was to carry out a series of experiments in the Royal Laboratory which proved that the grain was rotten and would not last, and that the method of proving gunpowder by a vertical eprouvette was not valid. At the same time, as a practical experiment, Congreve collected barrels of every type of gunpowder that he could find and sent it out to the West Indies in the 50-gun 4th Rate "Grampus". The result was vindication of his experiments in the Laboratory: all the powders he sent were examined before dispatch and found to be good; when re-examined after the voyage both the Government and the merchant powders were all spoiled. The merchant powder had formed into unusable lumps varying from 27% to 90% of the barrel. The result would have been in practice a serviceability rate of about 60%. By contrast, the foreign powders, Dutch, Swedish, and Russian, were all perfectly dry, separate and serviceable."
"At the same time, on Congreve's orders, all the powder in the Navy was returned to store and re-examined. It was said at the time that in some major ships there were only ten barrels of powder fit for use. In fact, the examination showed a 64% serviceability rate: not as bad as had been thought, but bad enough, and confirming both the Laboratory and "Grampus" experiments. The last series of experiments had been in 1754 (T. Fortune, "The Artillerist's Companion", 1778, Millan), when out of eleven powders tested comparatively, English cannon powder came eighth. The position seemed to have worsened."
"This was enough to refute the powder makers' claim that as things
stood they could make good powder. Congreve then set about a further
series of experiments in the Royal Laboratory to improve the situation.
He started making gunpowder himself. The then current proportions were
75% saltpetre, 15% charcoal and 10% sulphur. Congreve tried this, a
75,15, 9 mix and various other recipes including the standard French,
Swedish, Hanoverian and Dutch powders. His conclusion was that all the
systems of proportion dealt to the same standard, the 75,15,10 mix was
best. He then started experimenting with methods of manufacture,
particularly grain size, shape and durability."
"In the course of his investigations, Congreve spoke to one of the
powder makers, John Munns, and asked him why his powder grains had
degenerated virtually immediately after passing proof. Munns replied
that he had not made it to keep, merely to make it stong. Munns had a
point: from the manufacturer's point of view, provided the powder
passed proof, he was paid for it, and that was the end of the matter;
what happened thereafter was not his concern. However, it was very much the concern of the Board of Ordnance: this lack of durability was one of the major problems. Having experimented further, Congreve found that the fault lay in the manufacture. When powder was made it was pressed into cake, and then broken up into grains. The crucial factor was the pressure used to form it into cake: If too little pressure was used,
the powder could be grained, but it did not achieve homogeneity, and
thus began to degenerate almost immediately. What was needed was an
improved method of manufacture, and this could only be achieved by the
Board having a greater rather than a lesser part in powder making."
"Congreve also found that various woods in charcoal had different
characteristics, and concluded that dogwood was best, followed by
willow. In 1785 he had a considerable correspondence on the subject
with Richard Watson, who recommended that the charcoal be prepared in
cylinders, a system invented by a Dr. Fordyce and developed by Watson.
Following comparative experiments with this "cylinder charcoal",
standard pit charcoal, and an alternative variety peculiar to the powder
makers of Battle in Sussex, in which the charcoal was charred in pots,
Congreve confirmed that Watson's cylinder charcoal was best."
"While the durability of the grain was clearly of major importance,
Congreve found that size was also a crucial factor; he seems to have
been influenced by Benjamin Thompson (subsequently Count Rumsford) and J. Ingenhousz. Ingenhousz read a paper before the Royal Society in 1779 in which he established the precise role of each of the three constituents of gunpowder, and proved that grain size was the controlling factor in the size of inflammation of powder, from which it followed that grain size had to be proportional to the size of the gun. Thompson established in 1781 ("New Experiments of Gunpowder", Phil. Trans. Royal Society, 1781), that not only did compression have a major effect on powder quality, but that both gun temperature and charge temperature were also key factors, all of which had major implications for the proof system. Congreve established for himself that grain shape also had an effect: that the area of burning surface relative to the mass of the grain affected the rate of inflammation, so that "flake" powder burned faster than spherical grain; however, predictability depended on the
shape of the flake being regular, a development which at the time could
not be consistently achieved."
"Thompson also proved that the current method of proof, involving the vertical eprouvette, were totally unreliable, resulting not only in bad
powder being passed but in good powder being rejected. This agreed
with both Robins' and Hutton's observations, and were confirmed by
Congreve's own experiments. An experiment in 1780 at Faversham had
suggested that the French method of proof, using an eprouvette mortar, a standard quantity of powder, and a standard projectile (PRO WO 47-96)
was superior, and this was the method established by Congreve.
Eprouvette mortars can still be seen in situ in England."