The bark Endeavor ran aground on the Australian Great Barrier Reef in 1770. Anchor was located and recovered in 1971. About 13% of the original weight had been lost by corrosion.
The bark Endeavor ran aground on the Australian Great Barrier Reef in 1770. Anchor was located and recovered in 1971. About 13% of the original weight had been lost by corrosion.
The bark Endeavor ran aground on the Australian Great Barrier Reef in 1770. Anchor was located and recovered in 1971. About 13% of the original weight had been lost by corrosion.
The Met al l ography of a Wrought Iron Anchor from the Bark
Endeavor L. E. SAMUELS Materials Research Laboratories, Defence Seience and Technology Organization, Ascot Vale, Victoria, Australia The met al l ography of a wrought iron anchor j et t i soned near t he Aust ral i an coast in 1770 and r ecover ed in 1971 is descri bed. Some concl usi ons are drawn about t he iron making and forging i ndust ry of t he mid-18th Cent ury. Introduction An accompanying paper [1] describes the metallography of several cast iron items that were jettisoned by Lieutenant James Cook, Royal Navy (more commonly known as Captain Cook), when the Bark Endeavour ran aground on the Australian Great Barrier Reef in 1770. Cook lost the smaller of his two bow anchors during his efforts to refloat the Endeavour, and this anchor was located and recovered in 1971. The anchor when recovered was covered with a thick layer of coral, which had protected it well from corrosion: this coral was removed mechanically, and the surface corrosion products so exposed were removed by electrochemical means. The generally excellent condition of the anchor after restoration is indicated in Fig. 1. Three of the four stock bands were recovered with the anchor but not the stock itself; the stock would have been made of wood and had dis- integrated. The restored anchor shown in Fig. 1 is fitted with a modern stock made to the standard dimensions known to have been used at the time, and with modern stock bands made to the dimensions of those re- covered with the anchor. The shank of the anchor was about 3.95 m long and the span of the arms was about 2.25 m. Its original weight was marked on the shank as being 17 hundredweight 3 quarters and 14 lb (909 kg); its weight after restoration was 804 kg so that about 13% of the original weight had been lost by corrosion during 200 years immersion in the sea. The anchor is quite a large forging by any standard, but not a large anchor for the time, since anchors weighing up to 4500 kg with a shank El sevi er Nor t h Hol l and, Inc. , 1980 0026-0800/80/040357 + 12501.75 52 Vander bi l t Ave. , New York, NY 10017 358 L. E. Samue l s FIG. 1. The anchor after restoration. The stock is a modern replacement, as are the four iron bands that clamp the two longitudinal halves of the stock together. These bands were made to the same dimensions as the original bands that were recovered with the anchor. The stock was made to the standardized design dimensions of the time. length of 6 m were made. It had certainly been manufactured before 1770, but how much earlier cannot be ascertained. It can be taken, however, to be an example of a major iron forging manufactured just before the advent of the Industrial Revolution. This anchor would have been fabricated from wrought iron made by the so-called Walloon process of refining cast iron. This process was in- troduced early in the 16th Century and dominated iron production until about 1784, when it was replaced by the puddling process developed by Henry Cort. In the Walloon process, the end of a pig of cast iron was melted progressively in a charcoal-fired hearth furnace (the f i nery) , the molten iron being allowed to trickle down past a strongly oxidizing air blast. A semisolid mass (the loup) accumulated in the hearth of the furnace under a slag, and this mass was then repeatedly exposed to the oxidizing blast in the furnace until judged to be properly refined. Silicon and then carbon were removed by preferential oxidation, but little other refining lron Anchor from the Endeavour 359 occurred. Consequent l y, the quality of the product was determined largely b~) that of the feed pig iron. The lumps of spongy iron finally obtained were reheat ed in a furnace known as the chafery and kneaded to express as much as possible of the ent rapped oxide and slag. It was then forged into a bar, which weighted only about 20 kg. The process is descri bed in detail in Refs. [2] and [3]. Up until 1784, when Cort ' s puddled iron became readily available, the British Admiralty required that its anchors be manufactured from a mix- ture of Spanish and Swedish irons. This was because British pig irons, although well suited to castings, were known not to produce the best product when refined by the Walioon process. Forty-five to fifty refined bars woul d have been needed to fabricate the anchor being di scussed here. Groups of bars had first to be forge wel ded together to form the main units (such as the shank, arms, or palms) and these units had then to be j oi ned also by forge welding. The British Admiralty fabricated an- FIG. 2. A palm of the anchor after restoration. The macrostructure has been developed during corrosion, and the individual bars from which the palm had been fabricated can be seen. Note the remnant s of a bar welded across the back edge of the palm. 360 L. E. Samuels chors in its own dockyards, a major anchor works being attached to the Port smout h Dockyard. A visitor there in the mid-18th Cent ury report ed that " s event y or eighty br awny fellows were amongst their fi res" [4]. Trip (or helve) hammers driven by a wat er wheel were available at the time, but t hey required a copi ous supply of wat er and were limited in power. At least some of the forging of this anchor must t herefore have been carried out by hand. Cont emporary report s refer to a hammer, known as " He r c ul e s , " used to forge the shank to the arms of anchors in which a drop weight was raised by a t eam of men hauling on a rope. The handling of the anchor during forging must have been no mean feat, also. Macrostructure of the Anchor Corrosi on of the anchor component s had occurred i nhomogeneousl y in such a way that the grain st ruct ~t ~had been devel oped as if the anchor had been macroet ched. The bars from different heats of iron could be recognized, a particularly good example being shown in Fig. 2, and the forge welds bet ween major segments of the anchor could also be located (Fig. 3) and so the sequence of fabrication could be established. All welds were of a simple butt or scarf type. The interesting feature of the fabrication, however, was the procedure used to attach the arms to the shank. This sequence, as determined from observat i on of the ma- crost ruct ure and from radiography, is illustrated in Fig. 4. A double-ta- F~c. 3. Two views of the throat region of the anchor after restoration. The macrostructure developed during corrosion enables the sequence used to attach the arms to the shank to be established. lron Anchor from the Endeavour 361 Fl 6. 4. Model illustrating the sequence used to attach the arms to the shank of the anchor. pered flat tang had been forged on the end of the shank, the flats being aligned at approximately 45 to the intended axis of the shank (labeled S in Fig. 4). A taper had been formed on one side of the end of each arm (A) and a scarf weld made between this tapered face and one side of the tang on the shank, one arm being welded to each side of the shank. A V- shaped gap was thereby left at each side, because the flat on the arm necessarily was longer than that on the shank, and this V was filled in two stages with V-shaped pieces of iron (Vt and V2 in Fig. 4); a number of small gaps remained, nonetheless. The result is not a strong joint either mechanically or metallurgically, and so it is not surprising that it is re- ported that anchors frequently failed in service at the junction between the arm and the shank (called the throat or trend). Improvements had to await the development of more powerful methods of forging, the invention of the steam hammer by Nasmyth in 1842. Even then, observations on other anchors indicate that the above general procedure was used at least until the end of the 19th Century but that the more powerful forging ham- mers permitted joints with a better mechanical configuration and welds with much larger areas of overlap to be made. The only other feature of interest was that a bar had been welded across the back edge of each of the palms, so covering the end grain of the palm; the remnant of this bar is visible in Fig. 2. This arrangement would have reduced the ingress of corrosion into the back edge of the palm, and it seems likely that this was the intent. Microstructures of the Wrought Iron Material sufficient for chemical analyses could be removed only from the nut, which is a rectangular block welded onto the side faces at the 362 L. E. Samuels end of the shank for positioning of the stock. The analysis of the material was" 0.10% C; 0.01% Si; 0.11% Mn; 0.005% S; 0.016% P; 0.005% N. The sulphur and phosphorus cont ent s are bot h very low for a wrought iron, and could have been obtained by the Walloon process only when the feed pig had correspondingly low sulphur and phosphorus cont ent s. It is known that English charcoal-fueled blast furnaces could produce irons with low sulphur cont ent s but, due to the nature of the ores available, even the best English pig irons had comparat i vel y high ( - 0. 6%) phos- phorus cont ent s [1, 2]. The availability of ores necessary for the produc- tion of irons with low phosphorus cont ent was the special advant age that the Spanish and Swedi sh industries had, although the metallurgical rea- sons certainly would not have been known at the time. The carbon cont ent of the iron was a little higher than that which woul d be found in a wrought iron produced by Cort ' s puddling process t and, moreover, the results report ed bel ow show that it was variable even over small vol umes of material. The silicon cont ent was much l ower than for a puddl ed iron, bot h because pig irons made in the charcoal furnaces of this time had low silicon cont ent s and because silicon was removed ef- fectively by the Walloon refining process; there was no need, of course, for silicon as a deoxidant. Manganese was also low, which was accept abl e when t hes ul phur cont ent was so low. The above anal yses are, as usual, average figures for a comparat i vel y large vol ume of material. A section about 1 cm square was available for mi croscopi cal examination and three bands of distinctly different struc- ture and carbon cont ent were present within this area. In order of in- creasing carbon cont ent , the structures of these regions were: (i) Ferrite of moderat e grain size with some thick films of cementite at the grain boundari es, this cementite probabl y being a degenerate form of pearlite [Figs. 5(a) and (b)]; a moderat e number of globular nonmetallic inclusions present; hardness 77 HV. (ii) Ferrite of moderat e grain size with films of cementite at the grain boundari es and a small vol ume fracture of a degenerate pearlite [Figs. 5(c) and (d)]; a large number of elongated duplex inclusions present; hardness 78 HV. (iii) Ferrite of smaller grain size with a larger vol ume fracture of some- what degenerat e pearlite [Figs. 5(e) and (f)]; very few nonmetallic inclusions present; hardness 80 HV. A good-quality puddl ed iron would analyze: 0.08% C, 0.10% Si, <0. 10% Mn, 0.05% S, and 0.04% P[5]. " U O ! ~ J u o q a ~ a l e ! p a t u J a l u I ( 3 ) ' U O l ~ a a u o q a ~ ~ o q ( q ) ' H O l l O a u o q J ~ ~ o q ( ~ ) ( ' l e l ! u u t p a q 3 1 ~ ) a o q ~ u ~ ~ q l J o l n u a q l t u o J j u ~ u ! ~ a d s ~ u ! s p u e q ~ a a q l j o s ~ J n l 3 n J l s o ~ ! ~ " ~ " ~ t d ~ , , 4 m ~ - - ' + 7 - O O i ~ ~ " ' " ' m ~ O O t ' " ' " ' " + + + I + " . f t p J H + 6 " + , ,
/ v I P I x " l P m ~ o o t + t + . + . p t I I " \ \ t ~ 3 Z z . u a v a F u J a ~ l l ~ u . + j z v q J u V u u z I 364 L. E. Samuels The second structure corresponds to a carbon content of about 0.1%, the first to a slightly lower carbon content, and the third to a somewhat higher carbon content (about 0.2%). The structures are all characteristic of material that has been slowly cooled from above the A3 temperature (see Fig. I1 in Ref. 6); degenerate growth of pearlite is then common in low-carbon steels and in this instance can be attributed primarily to the low manganese content. The hardness of the material in all three regions was lower than that which would be found in a modern steel of the same carbon content and structure, and also lower than would be expected in a puddled/wrought iron. The low hardness can be attributed to the low levels of Si, Mn, and P, and indicates that the material would be of low strength. The two phases in the nonmetallic inclusions, as indicated by qualitative electron probe microanalysis, were iron oxide [light areas in Fig. 6(a)] and a complex iron-calcium-aluminum silicate [dark background in Fig. 6(a)]. Small amounts of manganese and magnesium were also present in both phases. This is about what is to be expected from the compositions of typical finery slags found at archaeological sites [2], so the inclusions can be taken to be pieces of finery slag that were not expelled in the chaffery. However, large inclusions of the type illustrated in Figs. 5(c) and 6(a) were rare in specimens examined throughout the anchor; in gen- eral, expulsion of oxide and slag had been efficiently carried out in the chaffery. General etch pitting developed in the ferrite grains in all three bands of material [Fig. 5(d)2], this pitting being characteristic of a low carbon steel in the quench-aged condition (see Fig. 15 in Ref. 6). From this, the following conclusions can be drawn about the final steps of the thermal cycle to which the material had been subjected. First, as a penultimate step, it must have been cooled fairly rapidly from about 650-700C to retain carbon (and nitrogen) in solution, which is difficult to reconcile with the conclusion reached earlier that the transformation of austenite which immediately preceeded this step had occurred during compara- tively slow cooling. Perhaps the forging was quenched after it had been cooled partly in air after each forging stage. Second, as a final stage, the material must have been overaged considerably to cause the degree of etch pitting observed. If this ageing occurred during manufacture, some tens of minutes at 150C would, for example, have been required, which presumably could have occurred only during later forging of a remote 2 This photomicrograph has been printed more darkly than the remainder of the series, better to show the etch pitting, but similar etch pitting was present in all of the areas illus- trated. lron Anchor f rom the Endeavour 365 (a) ~' ,~ 2 0 / ~ m (b) { i ; \ , f ~ J / - / 4 I / 2 0 ~ m o w. ~' ~ ~- ( c ) - " /~+. '~ " ' ~ ' + ~ 1 O 0 ~ m FiG. 6. Microstructures of different regions of the anchor. (d) J Etched in nital.) (a) Typical large inclusion in the nut [cf. Fig. 5(c)]. (b) A region of high nitrogen content in the palm. (c) A cold worked region in the palm. (d) As for Fig. 5(c). area of the anchor; this is unlikely because, as shall later be shown, the ferrite in all regions of the anchor showed evidence of similar overageing. Alternatively, it is a real possibility that 200 years immersion in tropical wat ers at 20-25C could have caused the required degree of ageing. Apart from the sample from the nut j ust di scussed, samples for micro- scopical examination could be obtained only by breaking off small por- tions of the fins left bet ween the more deepl y corroded areas, and even this was possi bl e only for the palms Most of the samples so examined had mi crost ruct ures within the range illustrated in Fig. 5, the majority having st ruct ures similar to that illustrated in Figs. 5(a) and (b), although some regions were rather more coarsel y grained. Any pearlite present 366 L. E. Samuels al ways was degenerat e, presumabl y indicating that a low manganese con- tent was general. Et ch pitting due to quench aging could be devel oped in all areas, although to varying degrees. Most specimens contained only a few spherical nonmetallic inclusions. However , unusual structures were obser ved in several regions. About 10% of the samples examined contained plates of precipitated nitrides, an advanced exampl e being shown in Fig. 6(b). These precipitates are indicative of a high nitrogen content, analysis of the specimen illustrated in Fig. 6(b) indicating a nitrogen cont ent of 0.018%. Nitrogen cont ent s of these levels are known to induce marked embrittlement. They also result in a considerable increase in hardness, as the hardness values in the caption to Fig. 6 indicate. The absorpt i on of nitrogen might be ex- pect ed with the Walloon refining processes, which leads to the suspicion that the material generally had a high nitrogen content. However , several samples which did not contain nitride precipitates were analyzed and found to contain nitrogen in the range 0.002-0.007%. Thus it appears that the nitrogen cont ent of Walloon irons, although variable and occasionally high, was frequent l y of the same order as that found in modern steels. One region examined had been cold worked, having recei ved a cold reduct i on of about 25% [Fig. 6(a)], which resulted in a significant increase in hardness. It is report ed that anchors were finished by cold hammering in the bel i ef that this i mproved corrosi on resistance, in which event only a surface layer probabl y would have been cold worked. The first possi- bility t herefore is that the material represent ed by Figs. 6(a) and (b) is the remanant from such a surface layer, the col d-worked layer having been removed from most ot her areas by corrosi on (loss in weight indicates that an average thickness of 3 mm was lost by corrosion). The second pos- sibility is that the forging of this particular area inadvertantly was finished at an excessi vel y low temperature. On the whole, the latter seems to be the more likely of the two. Conclusions The quality of the wrought iron in this anchor was inferior to that pro- duced by the later puddling process, essentially in that it was more het- erogenous. It generally had a slightly higher carbon content, was prone to have a higher nitrogen content, and contained a smaller vol ume fraction of nonmetallic inclusions. When made from appropriate pig irons, as in the present instance, refined irons with particularly low manganese, sil- icon, sulphur, and phosphorus cont ent s could be produced, iron which would have had poor strength but good corrosi on resistance. Iron Anchor from the Endeavour 367 The first skill in quality control required of a Walloon iron maker was the selection of feed pig iron. Thereafter, the only parameter over which he had control was carbon content, which would have been determined by the stage at which the repeated decarburization treatments was ter- minated. This point must have been determined by the visual judgment of the operator, as indeed it still was with the later puddling process [5]. It seems from the evidence obtained in the present examination that Walloon iron refineries could reduce the carbon content to below about 0.2% consistently, which is a remarkable enough achievement and raises the question of whether any quality tests were applied to the final product. It is known that iron puddlers used a fracture test to sort out irons that had not been adequately decarburized [5], and it would be interesting to know whether Walloon iron makers did likewise. This would be an im- portant step in tracing back the applications of metallography. The quality of these Walloon irons was certainly a considerable advance over that of irons produced by the earlier direct-reduction processes, the direct-reduced irons being much more heterogenous; they often contained regions with quite high carbon contents. For example, the structures of a number of nails made from direct-reduced iron produced somewhat before A.D. 87 have been described by Angus et al. [7]. These irons were much more heterogenous than those described here; the carbon content varying from 0.1-0.8% over distances of the order of millimeters. Many more large angular nonmetallic inclusions also were present in these nails. Direct-reduced irons were also prone to high nitrogen contents, some of the nails examined by Angus et al. contained nitride precipites to an even more marked extent than the irons described here. One of the saving graces of these early iron-making processes may well have been the very fact that the individual heats were so small. The final product was, in a sense, a composite which would have mitigated the deleterious effects of those individual heats that had inferior properties. The limitations of the Wallon process were that such small batches were produced and that the sources of high-quality iron were limited. The cost of the iron used in this anchor must have been very high in England, and an indication of the emphasis placed on quality in critical applications. This anchor did, after all, survive the trauma of hauling the Endeavour off of a reef. The anchor also gives an indication of the strengths and limitations of iron forging in the mid-18th Century. Considerable skill and brute man- power obviously was required to shape the components of this anchor, and to make so many sound forge welds. However, the procedure that had to be used to attach the arms to the shank is a clear indication of the limitations of the forging hammers that were available at the time. 368 L. E. Samue l s The as s i s t ance o f Wal t er He mmy , whos e wor k on t he det ai l s o f t he me t h o d by whi ch t he ar ms o f t he anchor were f o r g e d to t he shank, is grat ef ul l y ac k nowl e dge d. Ref erences 1. L. E. Samuels. The Metallography of cast iron relics from the Bark Endeavour, Metal- Iography 13: 345- 355 (1980). 2. R. F. Tylecote. A History of Metallurgy, The Metals Society, London (1976). 3. Leslie Aitchison. A History of Metals, MacDonald and Evans, London (1960). 4. A History of Technology, Vol. IV, (C. Singer, E. J. Holmyard, A. R. Hall and Trevor I. Williams, Eds. Clarendon Press, Oxford (1958). 5. H. D. Ward. ' Best Yorkshire' from West Yorkshire, J. Iron Steel Aust. 210: 396-405, (1972). 6. L. E. Samuels. Optical Microscopy of Carbon Steels, American Society for Metals, Metals Park (1980). 7. N. S. Angus, G. T. Brown, and H. F. Cleere, The iron nails from the Roman legionary fortress at Inchtuthil, Perthshire, J. Iron Steel lnst 200:956-968 (1962). Received April 1980.