The steel line 1850-1900: castings, rolled sections, plate, skelp and tube

Books The Manufacture and Properties of Iron and Steel, The Manufacture of Iron, The Manufacture of Iron and Steel Tubes, Roll-Turning for Sections in Steel and Iron, The Iron-Founder, An Autobiography, Treatise on Mills and Millwork, Part II

Not exhaustive. Covers the digests in this directory (marks, spencer, campbell, fairbairn, bolland, bessemer, open-hearth-books, books-synthesis-2, hamilton, overman) read by range, plus Campbell pp. 409-413 and Overman pp. 360-369 read from the books themselves (Campbell folio = PDF page less 32, Overman folio = PDF page less 6), plus online texts fetched 2026-10-25. Online texts are marked with their year; a page of Wikipedia or a company history is a secondary source and is said so. Page numbers are printed folios; "para" is a paragraph number where a text prints none. Purpose the period facts behind a steel line. Sections say what a source states; the lines marked Inferred are the note's own. Source dates: Greenwood, Steel and Iron, 2nd ed. 1884 (archive.org steelandironcom01greegoog); Campbell 4th ed. 1907; Washburn, Steel Castings, Univ. Illinois thesis 1906 (steelcastings00wash); Carr, Open Hearth Steel Castings 1907 (openhearthsteelc00carruoft); "The Rolling Mill Industry" 1913 (rollingmillindus00kindrich; author not read; cited as Rolling Mill Industry 1913); Encyclopaedia Britannica 11th ed. 1911 "Iron and Steel" on Wikisource. Several are after 1900; each such fact says so.

Answers in one place

  • Steel was cast from 1850 (crucible, Bochum bells) and in bulk from the 1880s (small open hearth, side-blown converters); it never replaced cast iron for machine frames before 1900 except in parts that took shock (hammer standards, tups, locomotive frames from 1893). Cast steel was a "retail" trade (EB 1911).
  • Rolled plate and sections riveted into girders, boiler shells and tubes (Britannia 1850, Forth 1890) are older than steel and carried straight over to it. Welded fabrication (forge-welded boiler parts) existed in 1884; fusion welding of structures is after 1900.
  • Wide plate: roughing and finishing pairs, rolls 20-36 in diameter (1884), 28-44 in (1913), armour 44-50 in; the Lauth three-high and reversing two-high both stand in the period. Wide flange (Grey) is 1902/1907, outside the main span.
  • Tube: the bend, the weld and the size were separate machines in the period (bending rolls, reheat, welding rolls on a mandrel, sizing rolls, cross-roll straightener, hydraulic test). A separate "bend and weld" machine is period.

1. Steel castings

Stated

  • Definition and processes (Campbell p. 409-410, 1907): a steel casting is "steel cast in a fluid state into the desired shape"; "must be made by pouring into finished shape the melted product of a crucible, a Bessemer converter, or an open-hearth furnace". Cupola-melted pig and steel is "a hybrid metal"; malleable iron is a different thing and is "extensively employed in making small, thin, or complicated shapes that could scarcely be poured in steel" (p. 409-410).
  • Crucible: "sometimes employed for small castings, since the conditions of the 'dead-melt' give a more quiet metal ... more apt to be free from blow-holes"; at Krupp's for big guns; "its great cost prohibits its adoption for general structural work" (Campbell p. 410).
  • Small converters (Campbell pp. 410-412): side-blown, loss "about 20 per cent, and in one case 30 per cent" against 10 per cent in standard practice (p. 411); "in 1903 there was not a single converter of [Clapp-Griffiths] type at work in America. Of the Roberts-Bessemer plants only two were active ... eight Tropenas plants ... All these plants were making steel castings" (p. 412). "The open-hearth furnace is the recognized agent for the making of steel castings"; most have acid lining; "currently supposed that the open-hearth furnace cannot make steel hot enough for small castings. This is a mistake" (p. 412).
  • Blow-holes and pipe (Campbell pp. 412-413): "no rapid or royal road to the production of sound steel castings" (Hadfield, quoted, 1890); aluminium three times and silicon four times as effective as manganese at deoxidising; cavities at the cut-off riser are "not evidence of unsoundness, but exactly the opposite" (feeding continued). Specification: phosphorus not over .04 per cent (p. 413). Digest: the Campbell note.
  • First dates (online): Bochumer Verein credits Jacob Mayer with "the invention of steel casting in 1850" (thyssenkrupp chronology, company history, fetched 2026-10-25); Hibberts (bell history, page unreachable on fetch, seen in search snippet only) puts the German development in the 1840s and the Naylor Vickers (Sheffield) licence in 1854. Cast-steel bells were shown against Krupp's claim at the 1855 Paris exhibition (campaners.com, snippet only). Krupp had a cast steel ingot at London 1851 (Wikipedia Krupp, secondary).
  • Bessemer castings: "the first attempt to make steel castings by the Bessemer Process was made in Upper Silesia in the early seventies, thin steel plates for street paving purposes being cast direct from the converter"; German works took the Walrand-Legenisel small converter in 1890 and Carl Rett began casting in 1896 (Washburn 1906, "Bessemer process" section, p. about 32 of the thesis; page not printed in the scan).
  • Open hearth: Heath conceived the plan 1845; Siemens regenerative furnace about 1860; Martin 1864 (EB 1911 Iron and Steel, para 98-99). Martin licensed from Siemens 1865 (Wikipedia Open hearth furnace, secondary).
  • What was cast, 1884: hammer standards "cast-iron ... or, as is now becoming more general, ... replaced by steel castings"; the tup "formerly always made of cast-iron, but is now frequently replaced by a steel casting" (Greenwood 1884, para 645, pp. 313-314). Steel moulds: the details "are very carefully kept secret"; moulds are "always well dried in the stove" with calcined ground quartz plus 2-3 per cent glue, water and flour, faced with fine quartz and graphite; Pourcel's silicon-manganese pig for sound castings (Greenwood para 872, p. 518 and para 659).
  • What was cast, 1893: Standard Steel Casting Co (Thurlow, Pa.; incorporated 22 July 1883, producing 1 March 1884) made "a complete cast steel frame for a locomotive" for Baldwin in 1893 (Wikipedia, secondary). Cast steel beds in one piece "an evolution of the bar frame ... in the latter years of steam locomotive design" (Wikipedia Locomotive frame, secondary); one-piece bed with cylinders 1926 per a Nashville steam page (single source, snippet).
  • 1906 range (Washburn): cast steel replaced forgings and iron castings in locomotive and car work, "a reduction in weight of about four tons ... on a passenger locomotive"; electrical machinery (high permeability); anvils, hammers, ingots from the Tropenas converter.
  • Fairbairn 1863 (digest, pp. 258-260): rolls are cast iron; "Staffordshire wire rolls are cast steel".
  • Limits against iron:
    • Melting point: "nearly 1500 C" (EB 1911, para 124, "These technical difficulties are due chiefly to the very high melting point"); "about 1,800 C (3,272 F)" (Greenwood 1884, p. 387). The two disagree; EB is later.
    • Contraction: "Steel in cooling contracts much more than gray iron cast from a similar temperature" (Carr 1907 p. 99); EB 1911 para 124 "large contraction", thin parts cool faster than thick ones so contract unevenly and tear or pipe. Greenwood 1884 (p. 387) says the opposite, steel moulds "a little less allowed for shrinkage" than iron, from the lower expansion coefficient. Carr is newer and describes steel foundry practice; no allowance in inches per foot was found in any source read.
    • Fluidity and temperature: "very hot and fluid as it is poured" (Washburn p. 17); steel is poured "at higher ranges of temperature than gray iron or malleable castings" (Carr p. 13); a large heat poured into many moulds "will lose its temperature" (Washburn, Carr p. 20); 93 per cent ladle yield is "fair average" (Carr p. 71).
    • Sand: "Pure silica is the most desirable", sharp irregular grains, refractory fire clay (Carr p. 13; Washburn).
    • Feeders: sink heads "heavy enough to act as an applied pressure"; "about forty percent of the metal tapped from the furnace is used in these sink heads" (Washburn 1906, annealing preface; the figure is the thesis's, not tied to a part). Larger headers than for iron (Carr p. 94). Contrast iron: a feeding head keeps hot iron against the central plane (Bolland digest, pp. 72-75).
    • Cracks: Carr p. 99-100: a cast needs fillets, no hard core or flask bar, because at the viscous point steel "will crumble under pressure".
    • Annealing: "From a theoretical standpoint all steel castings should be annealed" (Carr p. 101); as cast the steel is coarse and brittle, "snaps or fails suddenly on shock" (Carr p. 110); heated to the recalescence point "W" (800 C, 1472 F) and air cooled "an ideal structure" (Carr p. 109-110, Fig. 19); a 24 h soak and furnace cool also listed (temperatures in the OCR read "Fahr." and look wrong; not relied on). Greenwood 1884 (p. 388): annealing is "practised upon steel castings, forgings".
    • Cost and trade (EB 1911 para 124): "it is in fact much the more expensive"; each shape needs a new mould "often made largely by hand"; "a detail, one is tempted to say a retail, manufacture"; castings "are not so good as rolled products"; thin sections such as rails, plates, beams cannot be cast directly.
    • Failures: the U.S. government returned to wrought iron anchors from cast steel kedge anchors (Washburn 1906).
  • What stayed cast iron (stated): Fairbairn 1863 digest (millwork, housings, rolls, shafts cast); Bolland digest (cast beams, lintels, girders loaded across; cylinders; pipes; hydraulic cylinders; roll iron); Greenwood 1884 (rail rolls and mill housings cast iron, plate rolls chilled iron; para 568); Rolling Mill Industry 1913 (cast pipe made "in foundries specially built", under 100 psi). Steel castings in the U.S. 1910: 940,832 gross tons against 27.3 million of pig iron (Rolling Mill Industry 1913, statistical section; the OCR leaves the columns unheaded, the first matches 1910 totals).
  • Ruled: Bessemer steel was held unfit for structural castings and bridge tension members in the 1880s (see 2).

Inferred

  • The cast-iron survivors at 1850-1900 were machine frames, beds, housings and cylinders: the sources show steel castings entering where iron failed under shock (tups, standards, locomotive frames, anchors, wheels), not where iron was cheap and in compression.
  • A steel cast frame is a late (1890s-1900s) small-converter or open-hearth product with dried quartz moulds, an annealing step and a heavy head.
  • A rolled-and-riveted frame needs no foundry station.

2. What steel replaced; built-up against cast frames; welding

Stated

  • 1850: the Britannia tubes were "riveted wrought iron plates", 5/8 in thick, first rivet 10 Aug 1847, last 5 Mar 1850 (Wikipedia Britannia Bridge, secondary); Fairbairn tested the tube shape.
  • 1874: Eads Bridge "pioneered the large-scale use of steel as a structural material" with cast chromium steel components, wrought iron for much of the rest (Wikipedia, secondary); Carnegie supplied steel (Wikipedia steel history).
  • 1890: the Forth Bridge, 55,000 tonnes of steel (Siemens' Landore works 12,000 t of the original order, Steel Co. of Scotland, Dalzell), 6.5 million rivets, plates for the tubular members "heated, pressed to shape, assembled on a mandrel, and drilled for rivets"; began 1882, opened 4 Mar 1890 (Wikipedia Forth Bridge, secondary).
  • Steel rails replaced iron by 1883 in new and replacement track (Potter, Construction Physics, "Rise of Steel II", 2020s blog). Bessemer share of U.S. steel 88 per cent 1875, 66 per cent 1900, with open hearth 33 per cent in 1900, 66 per cent in 1911 (same).
  • 1887: "Not a few experienced engineers are now stipulating ... that Bessemer steel will not be allowed"; Bessemer beams "snap in two" (Potter, quoting period engineers); open hearth for railway bridges, boilers and locomotive forgings (Campbell pp. 24-25). Steel columns and building frames: Wermiel's 1890-91 date (search snippet only; paper not read).
  • Bessemer plate "widely used for ship plate" by 1870 (Wikipedia steel history, secondary; no primary found).
  • Welded fabrication: in 1884 "large front plates of marine and other boilers ... longitudinal flues and cross-tubes, are now frequently so welded" (forge welding of soft steel, Greenwood p. 387). Arc welding: carbon arc patents 1885 (UK) and 1887 (US), resistance welding Thomson 1885, covered electrode 1907-14, welding "did not become practical" until about 1900 (TWI history, fetched 2026-10-25). Riveted pipe is "generally of diameters larger than 30 inches" (Rolling Mill Industry 1913, p. 51).
  • Machine frames: Bolland digest (lathe beds and engine foundations single castings; fabricated frames in ribbed castings bolted to panels); Greenwood 1884 (hammer girders "of wrought-iron plates and angle irons riveted"; cast standards).
  • Locomotive bar frames (1830s) and plate frames were built up from rolled stock; cast steel beds came in the 1920s (Wikipedia Locomotive frame, secondary).

Inferred

  • Steel frames were rolled plates and sections riveted until the 1890s; a cast steel frame is a post-1893 option and for small parts.

3. Section rolling

Stated

  • I rail, iron, 1851 (Overman pp. 360-362): the square billet takes six grooves, No. 1 cut entirely in the bottom roller, Nos. 2-3 work the flanges, No. 4 smooths and thins the bottom flange, Nos. 5 and 6 "of the form of the rail when finished", "decrement ... very limited". Piles of rail iron are either welded under the T hammer or pushed through the roughing rollers to a six or seven inch billet and returned to the reheating furnace (p. 362). Saw cut rails to length (p. 363).
  • Steel rails, 1884 (Greenwood paras 577-579): six roughing and five finishing passes, a double length in a little over two minutes; U.S. 14 in. square Bessemer ingots of about a ton through a three-high cogging mill at 40-45 rpm, 11-13 passes in 1.5-2 min to a 7 in. bloom, then 7-9 passes in a 21-24 in. train at 75 rpm without reheating, a 90 ft rail cut in three by a hot saw; three to four minutes from reheat to rail bank.
  • Mills (Spencer digest, 1894): three-high roughing with two-high finishing, 8-18 in. mills; billet grooves 0.82-0.89 of the last; collar tapers for dressing; spread allowed. Mills for beams, 1913: "three-high with 28 to 36 in. rolls for heavy beams and channels 12 inches and over"; shape mills 20-26 in.; rail mills 26-40 in.; heavier shapes "frequently rolled direct in one heat from the ingot" (Rolling Mill Industry 1913, pp. 13, 17-19).
  • Reversing against three-high: reversing two-high mills stop after each pass and need heavy engines; three-high keep rolling and allow a heavy flywheel (Rolling Mill Industry 1913, pp. 11-12; Greenwood paras 562-565). Ramsbottom reversed the engine itself (Greenwood para 563, "Ramsbottom reversing mill engine"). EB 1911 para 129 agrees.
  • First rolled I-beams: France, with channels; Britain the first rails, angles, T and Z; Germany the first plates; U.S. the segmental shapes (Rolling Mill Industry 1913, p. 1-2, undated). Halbou, Forges de la Providence, 1849 (Wikipedia I-beam, search summary); Zores and a Paris carpenters' strike 1845 (search snippet of a Cambridge paper, 404 on fetch); U.S. rolled I-beams by Trenton and Phoenix in the 1850s (Hagley, Wermiel project page).
  • Universal mill (plain): "the so-called Universal or Belgian mill, for rolling plates and bars, has two horizontal rolls ... and a pair of vertical rolls working in front" so the edges are squared (Greenwood 1884, para 574, p. 336); universal plates 18 to 60 in. wide (Rolling Mill Industry 1913, p. 29). Daelen's design is dated 1840 by a Hoerde museum record (search snippet only).
  • Wide flange (Grey): "Grey developed his rolling process in 1897", Bethlehem 1907-08, the Differdingen works rolling similar sections from 1902; horizontal and vertical rolls form web and flange in one operation and a second set forms the flange tips (Bethlehem catalogue 1911 via search snippet; informedinfrastructure blog; steelexplained blog). Dates disagree by a year (1896/1897; 1907/1908).

Not found: pass count and groove sequence of an I-beam or channel; the stock a beam mill began from in 1850-1880 (the 1913 survey says ingot direct, blooms for smaller shapes); a groove drawing of a Zores beam.

4. Plate mills

Stated

  • Wrought iron boiler plate, 1851 (Overman pp. 366-369): mill bars from piles; slab from the T hammer "twelve to eighteen inches long, seven to ten inches wide, two to three inches thick"; one pair of rollers, rolled "in one heat" at bright red but not welding heat; cold water sprinkled with a broom loosens the scale; size set by turning the slab; rollers four to five feet long; 20-40 rev/min; screws and pinions "advantageous" for boiler plate.
  • Plate mills, 1884 (Greenwood paras 568-571, pp. 331-334): two pairs of rolls, 20-36 in. diameter and 4-9 ft long; the roughing top roll is counterbalanced, roughing rolls are "grain rolls" of tough unchilled iron, finishing rolls chilled 1/2 to 1.5 in. deep; the finishing top roll "runs loose" turned only by friction of the plate; the gap is reduced "at each pass by screwing down the feed-screws"; reversing mills "most extensively adopted" for plate, with a horse and bogie; the top roll slightly larger so the plate curls down and does not collar; extension is almost all lengthwise, so the slab is first rolled across to the width, then turned and rolled lengthwise; the plate is then sheared. Speed 25-30 rpm for reversing plate mills with 20-36 in. rolls; steel mills run 25-30 per cent faster and need more power (para 576).
  • Fairbairn 1863 digest (pp. 259, 269-270): plate rolls over 6 ft by 18-21 in., 28-30 rpm; the armour-plate mill 32 in. by 8 ft rolls on a 400 hp engine, screw through compound levers each pass (12, 10, 8 in.).
  • EB 1911 para 130: armour plate rolled between plain cylinders; Krupp's mill rolls 4 ft diameter by 12 ft; as much as 40 per cent of the ingot may be rejected from piping (para 121).
  • Rolling Mill Industry 1913: plate mills 28-44 in., armour 44-50 in. with 140-180 in. body; "the maximum width which can be rolled fixes the size of the mill"; plates 12 gauge and thinner are sheets (pp. 12, 19, 29).
  • Lauth three-high: the middle roll is small and free, power on top and bottom (Danish heritage scan, a book page seen in a search; Holley, AIME 1873, cited by search only). The 1913 survey calls three-high "non-reversing". Wikipedia Rolling gives "introduction of three-high mills in 1853 used for rolling heavy sections".
  • Slab and ingot: slab "not less than 2 inches" thick and 12 in. wide (Rolling Mill Industry 1913, p. 20); the Spencer digest slab roll (Plate I) with a 90 in. barrel, 48 in. plain and 30 in. top lift. Bessemer crops remelted (Bessemer digest, p. 65).

Not found: Lauth's date and patent; the number of passes and heats for a steel plate; a crop allowance for a plate slab.

5. Skelp and tube

Stated

  • Butt weld (Marks digest, 1825): Whitehouse, flat "plough-plate iron" bent up, put "into a hollow fire", drawn by a chain on a draw bench through dies at welding heat, one end at a time, 2-8 ft; lap weld from 1829 (Marks pp. 9-11, 127); 1913: butt weld to 3 in. through a "bell-shaped die" (Rolling Mill Industry 1913, p. 52); Marks p. 124: not ordinarily over 2 in. for steam.
  • Lap weld (Scientific American, 27 Feb 1858): scarfing machine cuts the edges, skelp heated in a reverberatory furnace and bent by dies, reheated to welding heat, "a pointed ball" on a rod with fluted rolls welds the lap "without leaving any fin"; friction rollers straighten; American production from 1852 (McCully's machinery), introduced from England about 1845; sizes 1 to 8 in., new machinery 12 in. By 1913 lap weld to 30 in., through bending rolls, reheat, welding rolls on a mandrel, sizing rolls, cross-roll straightening, cooling, cutting, hydraulic test; over 30 in. from bent plate with the edges raised by a burner and welded in a machine; riveted pipe over 30 in. (Rolling Mill Industry 1913, pp. 51-53).
  • Skelp: "rolled from blooms in skelp mills"; edges sheared for large pipe; "scarfed skelp" has bevelled edges for lap weld; grooved skelp is rolled in a mill with grooves of plate width (Rolling Mill Industry 1913, pp. 29-30). Skelp width is the tube's girth plus the lap (Marks digest). The bevel rolls at the furnace mouth are Muckley 1891; the roll-and-trumpet Huggins 1886; Howard 1890 former and bell; Twer 1898 bell with rollers (Marks digest).
  • Sizes (Marks): butt weld to 1/8 in. bore, lap weld from 1 in.; steam tube one gauge thicker than water than gas; test 300 lb/sq in. for water and steam. Boiler tubes 1-4 in., No. 13-6 gauge (1913 survey).
  • Furnaces: "hollow fire" with blast (Whitehouse); "reverberatory" (Prosser 1840, Scientific American 1858); a gas or oil fired furnace is not documented in the tube sources read.
  • Seamless: Stirling 1854 cast hollow and extended (Marks pp. 24-25); Mannesmann patent filed 1885, granted 1886, first seamless steel tube rolled 1886, pilger rolling in the 1890s (Salzgitter AG history, company page); the works were united 16 July 1890 (same). 1913: billet pierced over a mandrel, rolled, cold-drawn with pickle and anneal after each draw.
  • Steel for tube: wrought iron through 1903, boiler tube in "a very mild steel" (Marks p. 121).

Inferred

  • Marks and the 1913 survey together show a separate machine for each step: the bend is its own set of rolls and the weld another, with a furnace between.

What stayed unsettled

  1. Lauth three-high plate mill: inventor, date and whether the 1853 Wikipedia date relates to it.
  2. Steel shrinkage allowance in inches per foot; the sources disagree (Greenwood 1884 less than iron, Carr 1907 more).
  3. Steel melting point: 1800 C (1884) against nearly 1500 C (1911).
  4. Annealing temperatures and times for steel castings (Carr's OCR unusable).
  5. First steel casting date: 1850 Bochum against the 1840s; Bessemer castings early 1870s; first steel locomotive frame 1893 rests on one Wikipedia page.
  6. Pass counts and groove sequences for an I-beam, channel, angle, tee; the stock they started from before 1880.
  7. Plate passes, heats and crops for steel plate; Landore plate and boiler plate dates.
  8. Tube furnaces beyond the hollow fire and reverberatory; the pass count of a skelp mill; Stirling's 1854 patent; Fretz-Moon continuous butt weld (after the span, not searched).
  9. When rolled steel beams replaced wrought iron beams (Wermiel's paper not read; Hagley page only).
  10. The Fullagar all-welded hull date: TWI places it in 1914-1918; not checked elsewhere.
  11. Pages that failed to fetch: Hibberts steel bells, Grace's Guide (certificate error), the 1911 Bethlehem catalogue (403), the Cambridge Mende paper (404).