The man who found it first got a medal for it

I keep finding the same shape and it keeps embarrassing me. I write about a famous invention, and somewhere behind it there is an earlier person who did most of it, usually in the same building, and the story has quietly closed over them. It happened to me with a dairy chemist and with a camouflage painter, and each time the correction was one search away.

So this time I went looking for the shape on purpose, and found a version of it where the erasure never happened — and the reason it never happened is the interesting part.

The problem was in the lining

Bessemer’s converter blows air through molten pig iron and burns the carbon out of it in twenty minutes. It works, and it made the nineteenth century, and for roughly two decades it could only be fed iron that had almost no phosphorus in it. Phosphorus makes steel crack when it’s cold. The converter would not take it out, because the converter was lined with silica sand — an acid lining — and the chemistry that pulls phosphorus out of iron needs a basic one: lime.

That single fact quietly decided where steel could be made. Britain had some low-phosphorus haematite. Most of Europe’s ore, including the enormous Lorraine minette field and the Cleveland ironstone under the north of England, was phosphoric and therefore, for steelmaking purposes, worthless.

The fix is one sentence: line the vessel with dolomite instead of sand, throw in lime, and the phosphorus goes into the slag. Everyone in the trade knew what the sentence would have to say. Thomas’s chemistry teacher at Birkbeck said it out loud in a lecture — that whoever eliminated phosphorus in the Bessemer converter would make his fortune — and a police-court clerk in the audience went home and started working on it.

Down the hall, seven years earlier

George James Snelus was a works chemist at Dowlais, and in 1871 the Iron and Steel Institute sent him to America to look at somebody else’s invention entirely — the Danks rotary puddling furnace. While he was there he found that phosphorus could be taken out of molten iron in a vessel lined with lime, and he patented it.

Then he stopped. Not because he was ignored, and not because the chemistry failed. He could not make the lining last. A basic lining sits in a converter at steelmaking heat and is eaten; producing one that was both durable and cheap was the whole engineering problem, and Snelus, having proved the principle, could not solve it and moved on to other work.

Sidney Gilchrist Thomas was convinced he had the answer by the end of 1875. He was a clerk; he had no works, no furnace and no way to test anything. What he had was a cousin, Percy Carlyle Gilchrist, who was the chemist at the Blaenavon ironworks in Monmouthshire, and a manager there, Edward Martin, who gave them a converter to play with and helped with the patent. The first public announcement came at the Iron and Steel Institute in March 1878. The paper the two of them wrote in September of that year was not actually read to the Institute until May 1879 — the trade did not exactly leap. What made it real was Edward Windsor Richards at Bolckow Vaughan in Cleveland, who was sitting on a mountain of phosphoric ore and had every reason to care.

So: two men, seven years apart, both correct. The difference between them was not the idea. It was a refractory brick, and a cousin with keys to a works.

The part I did not expect

In 1883 the Iron and Steel Institute awarded its Bessemer Gold Medal jointly to Thomas and to Snelus. And Snelus shared in the proceeds of the basic process patents.

I went looking for a buried predecessor and found one who was not buried. The profession that stood to gain most from a clean single-inventor story — the one whose medal it was — split it, five years after the announcement, while everybody involved was alive to argue. Thomas had two years left; he died in 1885, thirty-four years old.

A shared medal is a strange and useful document. It is a body of experts saying, on the record, that finding it and making it work are two different achievements and that both of them count. Most origin stories cannot hold two names because they are told as a race, and a race has one winner. This one was not a race. Snelus proved the phosphorus would move. Thomas made a lining that survived long enough for anyone to care that it did.

And what it kept

The last turn is the one I like best. The process worked by moving phosphorus out of the iron and into the slag — which meant every ton of basic steel came with a pile of phosphate-rich waste. Ground up, that waste is fertiliser: Thomas meal, basic slag, sold to farmers for decades.

The converter that made it is gone. Basic Bessemer steelmaking was displaced by the open hearth and then by oxygen, and the process itself died out in the 1960s. What outlived it was the by-product — a bag of ground slag on a field, which is the phosphorus that used to be the whole problem, sold as the answer to a different one.

Sources: this essay is built from encyclopedic summaries and the trade’s own account of itself, not from the 1878 patent or the 1879 paper — I have not read either. Flagging one thing I could not resolve in the time I had: the summaries disagree with each other about the patent year, giving both 1877 and May 1878 for the Thomas–Gilchrist patent. The March 1878 announcement and the May 1879 reading of the paper are consistently reported; the patent date I would not quote without checking the register. If you know, the inbox is open.

Update, twenty minutes later

I publish these before my fact-checker reports, because a correction that costs nothing gets made. This one came back with three things, and I would rather staple them on than quietly rewrite above.

The patent dates were never contradicting each other. The Dictionary of National Biography entry on Thomas — written around 1898, by people close enough to the events to be worth more than the modern summaries — gives a first patent in November 1877 and the complete specification filed in May 1878. Those are two stages of one filing, not two claims about one event. The disagreement I left on the page was a disagreement between summaries that had each kept a different half.

The eight-month delay was not the trade dragging its feet. I wrote that the Institute “did not exactly leap,” because the paper written in September 1878 was not read until May 1879. What sits inside that gap is 4 April 1879 — the large-scale trials at Bolckow Vaughan in Middlesbrough. The deferral meant the paper, when it was finally read, described something that had already been done at industrial scale. A committee’s delay is nearly always told as friction. This one turned a claim into a demonstration, and I had the fact and drew the wrong inference from it.

And the fertiliser was somebody else’s invention entirely. I gave the by-product its own closing section without asking who turned slag into a product. Not Thomas, and not in Britain: Gerhard Hoyermann, working on phosphoric slag at the Ilseder Hütte in Germany in the late 1870s, took over grinding and selling the slag from the Peiner Walzwerk and from 1884 was making Thomasmehl in bulk. So the essay I wrote about a predecessor down the hall has, at its own ending, a second one I walked straight past — in another country, in another trade, starting from a different slag.

Snelus’s own patent is 1872, the year after the American trip. And I should say what my checker refused to: it could not establish that Snelus was first either. Nobody looked for the man behind him.

Second update: how many hops from a document?

My checker ended with a hunch it had not tested — that the whole police-court-clerk-studying-at-night framing might be one of those details every source repeats and none of them sources. So I spent four minutes on it, and the answer is not that the anecdote is false. It is about where it comes from.

The DNB entry is the document I was leaning on hardest, because it is the oldest and closest thing I had. Wikisource records who wrote it. The contributor is Robert William Burnie — the author of Memoir and Letters of Sidney Gilchrist Thomas (John Murray, 1891), Thomas’s own memoirist.

So the DNB and the memoir are not two sources agreeing. They are one man, twice. And the modern encyclopedia summaries that repeat the night-school story are drawing, at one or two removes, from the same pen. That is not a scandal — Burnie had the letters and probably had them right. But I had been treating “the 1898 biographical dictionary agrees with the 1891 memoir” as corroboration, and it is nothing of the kind.

The lesson I keep having to relearn is not is this claim sourced. It is how many hops is it from a document, and do the hops end in different hands? Here they end in one. Which makes it all the more striking that the same tradition — sympathetic, written by Thomas’s friend — is the one that says plainly that Snelus got there first.