William Kelly's Air-boiling Process
An account of Bessemer's address to the British Association was
published in the Scientific American on September 13, 1856. On
September 16, 1856, Martien filed application for a U.S. patent on his
furnace and Mushet for one on the application of his triple compound to
cast iron "purified or decarbonized by the action of air blown or
forced into ... its particles while it is in a molten ... state."
Mushet, by t
is time, had apparently decided to generalize the
application of his compound instead of citing its use in conjunction
with Martien's process, or, as he put it, he had been obliged to do for
his English specification by the Ebbw Vale Iron Works.
 Scientific American, 1856, vol. 12, p. 6.
 U.S. patent 17389, dated May 26, 1857. Martien's U.S. patent
was granted as 16690, dated February 24, 1857.
FURNACE, FROM U.S. PATENT 17628. A is "the flue to carry off the
carbonic gas formed in decarbonizing the iron," B is the port through
which the charge of fluid iron is received, C and C' are the
tuyeres, and D is the tap hole for letting out the refined metal.]
The discussion in the Scientific American, which was mostly concerned
with Martien's claim to priority, soon evoked a letter from William
Kelly. Writing under date of September 30, 1856, from the Suwanee Iron
Works, Eddyville, Kentucky, he claimed to have started "a series of
experiments" in November 1851 which had been witnessed by hundreds of
persons and "discussed amongst the ironmasters, etc., of this section,
all of whom are perfectly familiar with the whole principle ... as
discovered by me nearly five years ago." A number of English puddlers
had visited him to see his new process. "Several of them have since
returned to England and may have spoken of my invention there." Kelly
expected "shortly to have the invention perfected and bring it before
 Scientific American, 1856, vol. 12, p. 43, Kelly's
suggestion of piracy of his ideas was later enlarged upon by his
biographer John Newton Boucher, William Kelly: A true history
of the so-called Bessemer process, Greensburg, Pennsylvania,
Bessemer's application for an American patent was granted during the
week ending November 18, 1856, and Kelly began his interference
proceedings sometime before January 1857.
 Ibid., p. 82. Kelly's notice of his intention to take
testimony was addressed to Bessemer on January 12, 1857. See
papers on "Interference, William Kelly vs. Henry Bessemer
Decision April 13, 1857." U.S. Patent Office Records. Quotations
below are from this file, which is now permanently preserved in
the library of the U.S. Patent Office.
Kelly's witnesses were almost wholly from the ranks of employees or
former employees. The only exception was Dr. Alfred H. Champion, a
physician of Eddyville. Dr. Champion describes a meeting in the fall of
1851 with "two or three practical Ironmasters and others" at which
Kelly described his process and invited all present to see it in
operation. He stated:
The company present all differed in opinion from Mr. Kelly and
appealed to me as a chemist in confirmation of their doubts. I at
once decided that Mr. Kelly was correct in his Theory and then went
on to explain the received opinion of chemists a century ago on
this subject, and the present received opinion which was in direct
confirmation of the novel theory of Mr. Kelly. I also mentioned the
analogy of said Kelly's process in decarbonising iron to the
process of decarbonising blood in the human lungs.
The Doctor does not say, specifically, if he or any of the "company"
went to see the process in operation.
Kelly obtained affidavits from another seventeen witnesses. Ten of
these recorded their recollections of experiments conducted in 1847.
Five described the 1851 work. Two knew of or had seen both. One of the
last group was John B. Evans who became forge manager of Kelly's Union
Forge, a few miles from Suwanee. This evidence is of interest since a
man in his position should have been in a position to tell something
about the results of Kelly's operations in terms of usable metal.
Unfortunately, he limits himself to a comment on the metal which had
chilled around a tuyere which had been sent back to the Forge ("it was
partly malleable and partly refined pig-iron") and to an account of a
conversation with others who had worked some of Kelly's "good wrought
iron" made by the new process.
Only one of the witnesses (William Soden) makes a reference to the
phenomenon which is an accompaniment of the blowing of a converter: the
prolonged and violent emission of sparks and flames which startled
Bessemer in his first use of the process and which still provides
an exciting, if not awe-inspiring, interlude in a visit to a steel
mill. Soden refers, without much excitement, to a boiling commotion,
but the results of Kelly's "air-boiling" were, evidently, not such as
to impress the rest of those who claimed to have seen his furnace in
operation. Only five of the total of eighteen of the witnesses say that
they witnessed the operations. Soden, incidentally, knew of seven
different "air-boiling" furnaces, some with four and some with eight
tuyeres, but he also neglected to report on the use of the metal.
 Bessemer, op. cit. (footnote 7), p. 144.
As is well known, Kelly satisfied the Acting Commissioner that he had
"made this invention and showed it by drawings and experiment as early
as 1847," and he was awarded priority by the Acting Commissioner's
decision of April 13, 1857, and U.S. Patent 17628 was granted him as of
June 23, 1857. The Scientific American sympathized with Bessemer's
realization that his American patent was "of no more value to him than
so much waste paper" but took the opportunity of chastising Kelly for
his negligence in not securing a patent at a much earlier date and
complained of a patent system which did not require an inventor to make
known his discovery promptly. The journal advocated a "certain fixed
time" after which such an inventor "should not be allowed to subvert a
patent granted to another who has taken proper measures to put the
public in possession of the invention."
 Scientific American, 1857, vol. 12, p. 341.
Little authentic is known about Kelly's activities following the grant
of his patent. His biographer does not document his statements,
many of which appear to be based on the recollections of members of
Kelly's family, and it is difficult to reconcile some of them with what
few facts are available. Kelly's own account of his invention,
itself undated, asserts that he could "refine fifteen hundredweight of
metal in from five to ten minutes," his furnace "supplying a cheap
method of making run-out metal" so that "after trying it a few days we
entirely dispensed with the old and troublesome run-out fires."
This statement suggests that Kelly's method was intended to do just
this; and it is not without interest to note that several of his
witnesses in the Interference proceedings, refer to bringing the metal
"to nature," a term often used in connection with the finery furnace.
If this is so, his assumption that he had anticipated Bessemer was
based on a misapprehension of what the latter was intending to do, that
is, to make steel.
 Boucher, op. cit. (footnote 97).
 U.S. Bureau of the Census, Report on the manufacturers
of the United States at the tenth census (June 1, 1880) ...,
Manufacture of iron and steel, report prepared by James M.
Swank, special agent, Washington, 1883, p. 124. Mr. Swank was
secretary of the American Iron and Steel Association. This
material was included in his History of the manufacture of iron
in all ages, Philadelphia, 1892, p. 397.
 Ibid., p. 125. The run-out fire (or "finery" fire) was a
charcoal fire "into which pig-iron, having been melted and
partially refined in one fire, was run and further refined to
convert it to wrought iron by the Lancashire hearth process,"
according to A. K. Osborn, An encyclopaedia of the iron and
steel industry, New York, 1956.
This statement leaves the reader under the impression that the process
was in successful use. It is to be contrasted with the statement quoted
above (page 43), dated September 1856, when the process had, clearly,
not been perfected. In this connection, it should be noted that in the
report on the Suwanee Iron Works, included in The iron manufacturer's
guide, it is stated that "It is at this furnace that Mr. Kelly's
process for refining iron in the hearth has been most fully
 J. P. Lesley, op. cit. (footnote 39), p. 129. The preface
is dated April 6, 1859. The data was largely collected by Joseph
Lesley of Philadelphia, brother of the author, during a tour of
several months. Since Suwanee production is given for 44 weeks
only of 1857 (i.e., through November 4 or 5, 1857) it is
concluded that Lesley's visit was in the last few weeks of 1857.
A major financial crisis affected United States business in the fall of
1857. It began in the first week of October and by October 31 the
Economist (London) reported that the banks of the United States had
"almost universally suspended specie payment." Kelly was involved
in this crisis and his plant was closed down. According to Swank,
some experiments were made to adapt Kelly's process to need of rolling
mills at the Cambria Iron Works in 1857 and 1858, Kelly himself being
at Johnstown, at least in June 1858. That the experiments were not
particularly successful is suggested by the lack of any American
contributions to the correspondence in the English technical journals.
Kelly was not mentioned as having done more than interfere with
Bessemer's first patent application. The success of the latter in
obtaining patents in the United States in November 1856, covering
"the conversion of molten crude iron ... into steel or malleable iron,
without the use of fuel ..." also escaped the attention of both English
and American writers.
 Economist (London), 1857, vol. 15, pp. 1129, 1209.
 Swank, op. cit. (footnote 42), p. 125. John Fritz, in his
Autobiography (New York, 1912, p. 162), refers to experiments
during his time at Johnstown, i.e., between June 1854 and July
1860. The iron manufacturer's guide (see footnote 104) also
refers to Kelly's process as having "just been tried with great
success" at Cambria.
 U.S. patents 16082, dated November 11, 1856, and 16083,
dated November 18, 1856. Bessemer's unsuccessful application
corresponded with his British patent 2321, of 1855 (see footnote
It was not until 1861 that the question arose as to what happened to
Kelly's process. The occasion was the publication of an account of
Bessemer's paper at the Sheffield meeting of the (British) Society of
Mechanical Engineers on August 1, 1861. Accepting the evidence of "the
complete industrial success" of Bessemer's process, the Scientific
American asked: "Would not some of our enterprising manufacturers
make a good operation by getting hold of the [Kelly] patent and
starting the manufacture of steel in this country?"
 Scientific American, 1861, new ser., vol. 5, pp. 148-153.
There was no response to this rhetorical question, but a further
inquiry as to whether the Kelly patent "could be bought" elicited
a response from Kelly. Writing from Hammondsville, Ohio, Kelly
said, in part:
I would say that the New England states and New York would be sold
at a fair rate.... I removed from Kentucky about three years ago,
and now reside at New Salisbury about three miles from
Hammondsville and sixty miles from Pittsburg. Accept my thanks for
your kind efforts in endeavoring to draw the attention of the
community to the advantages of my process.
 Ibid., p. 310.
 Ibid., p. 343.
This letter suggests that the Kelly process had been dormant since
1858. Whether or not as a result of the publication of this letter,
interest was resumed in Kelly's experiments. Captain Eber Brock Ward of
Detroit and Z. S. Durfee of New Bedford, Massachusetts, obtained
control of Kelly's patent. Durfee himself went to England in the fall
of 1861 in an attempt to secure a license from Bessemer. He returned to
the United States in the early fall of 1862, assuming that he was the
only "citizen of the United States" who had even seen the Bessemer
 His claim is somewhat doubtful. Alexander Lyman Holley, who
was later to be responsible for the design of most of the first
Bessemer plants in the United States had been in England in 1859,
1860, and 1862. In view of his interest in ordnance and armor, it
is unlikely that Bessemer could have escaped his alert
observation. His first visit specifically in connection with the
Bessemer process appears to have been in 1863, but he is said to
have begun to interest financiers and ironmasters in the Bessemer
process after his visit in 1862 (Engineering, 1882, vol. 33, p.
In June, 1862, W. F. Durfee, a cousin of Z. S. Durfee, was asked by
Ward to report on Kelly's process. The report was unfavorable.
"The description of [the apparatus] used by Mr. Kelly at his abandoned
works in Kentucky satisfied me that it was not suited for an experiment
on so large a scale as was contemplated at Wyandotte [Detroit]." Since
it was "confidently expected that Z. S. Durfee would be successful in
his efforts to purchase [Bessemer's patents], it was thought only to be
anticipating the acquisition of property rights ... to use such of his
inventions as best suited the purpose in view."
 W. F. Durfee: "An account of the experimental steel works
at Wyandotte, Michigan," Transactions of the American Society of
Mechanical Engineers, 1884, vol. 6, p. 40 ff.
Thus the first "Bessemer" plant in the United States came into being
without benefit of a license and supported only by a patent "not
suited" for a large experiment. Kelly seems to have had no part in
these developments. They took some time to come to formation. Although
the converter was ready by September 1862, the blowing engine was not
completed until the spring of 1864 and the first "blow" successfully
made in 1864. It may be no more than a coincidence that the start of
production seems to have been impossible before the arrival in this
country of a young man, L. M. Hart, who had been trained in Bessemer
operations at the plant of the Jackson Brothers at St. Seurin (near
Bordeaux) France. The Jacksons had become Bessemer's partners in
respect of the French rights; and the recruitment of Hart suggests the
possibility that it was from this French source that Z. S. Durfee
obtained his initial technical data on the operation of the Bessemer
 Research in the French sources continues. The arrival of L.
M. Hart at Boston is recorded as of April 1, 1864, his ship being
the SS Africa out of Liverpool, England (Archives of the United
States, card index of passenger arrivals 1849-1891 list No. 39).
During the organization of the plant at Wyandotte, Kelly was called
back to Cambria, probably by Daniel J. Morrell, who, later, became a
partner with Ward and Z. S. Durfee in the formation of the Kelly
Pneumatic Process Company. We learn from John E. Fry, the
iron moulder who was assigned to help Kelly, that--
in 1862 Mr. Kelly returned to Johnstown for a crucial, and as it
turned out, a final series of experiments by him with a rotative
[Bessemer converter] made abroad and imported for his purpose.
This converter embodied in its materials and construction several
of Mr. Bessemer's patented factors, of which, up to the close of
Mr. Kelly's experiments above noted, he seemed to have no knowledge
or conception. And it was as late as on the occasion of his return
in 1862, to operate the experimental Bessemer converter, that he
first recognized, by its adoption, the necessity for or the
importance of any after treatment of, or additions required by the
blown metal to convert it into steel.
 Swank, op. cit. (footnote 42), p. 409.
 Johnstown Daily Democrat, souvenir edition, autumn 1894
(italics supplied). Mr. Fry was at the Cambria Iron Works from
1858 until after 1882.
Fry later asserted that Kelly's experiments in 1862 were simply
attempts to copy Bessemer's methods. (The possibility is under
investigation that the so-called "pioneer converter" now on loan to the
U.S. National Museum from the Bethlehem Steel Company, is the
converter referred to by Fry.)
 Engineering, 1896, vol. 61, p. 615.
William Kelly, in effect, disappeared from the record until 1871 when
he applied for an extension of his patent of June 23, 1857. The
application was opposed (by whom, the record does not state) on the
grounds that the invention was not novel when it was originally issued,
and that it would be against the public interest to extend its term.
The Commissioner ruled that, on the first question, it was settled
practice of the Patent Office not to reconsider former decisions on
questions of fact; the novelty of Kelly's invention had been
re-examined when the patent was reissued in November 1857. Testimony
showed that the patent was very valuable; and that Kelly "had been
untiring in his efforts to introduce it into use but the opposition of
iron manufacturers and the amount of capital required prevented him
from receiving anything from his patent until within very few years
past." Kelly's expenditures were shown to have amounted to $11,500,
whereas he had received only $2,400. Since no evidence was filed in
support of the public interest aspect of the case, the Commissioner
found no substantial reason for denying the extension; indeed "very few
patentees are able to present so strong grounds for extension as the
applicant in the case."
 See U.S. Patent Office, Decision of Commissioner of
Patents, dated June 15, 1871.
In a similar application in the previous year, Bessemer had failed to
win an extension of his U.S. patent 16082, of November 11, 1856, for
the sole reason that his British patent with which it had been made
co-terminal had duly expired at the end of its fourteen years of life,
and it would have been inequitable to give Bessemer protection in the
United States while British iron-masters were not under similar
restraint. But if it had not been for this consideration, Bessemer
"would be justly entitled to what he asks on this occasion." The
Commissioner observed: "It may be questioned whether [Bessemer]
was first to discover the principle upon which his process was founded.
But we owe its reduction to practice to his untiring industry and
perseverance, his superior skill and science and his great outlay."
 U.S. Patent Office, Decision of Commissioner of Patents
dated February 12, 1870.