Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5

A study of historical test data for better informed assessment ofwrought-iron structures

M. O’SullivanUniversity of Manchester, Manchester, UK

T. SwailesUniversity of Manchester, Manchester, UKSenior Engineer, Arup, UK

ABSTRACT: Wrought-iron was the dominant structural framing material from 1850 to 1890. With similarproperties to early mild steel, it is more variable, creating uncertainty in the assessment of existing structureswhere sampling and testing opportunities are limited. This paper considers the factors that determine howhistorical test data may be used to better inform the assessment of wrought-iron in structures. Strength values forwrought-iron lie between wide limits. Sources of variability include test methods used, the ‘grade’ or ‘quality’of wrought-iron, and the type of structural element tested or from which samples have been taken. Yield pointbecame of great importance when the use of elastic stress analysis for beam design became commonplace andby 1879 Watertown Arsenal in the USA was publishing stress-strain data. Ten years later testing wrought-ironwas a routine part of the education of engineers at universities.

1 INTRODUCTION

Over the course of the 19th century structural framingwas dominated first by cast iron, then wrought-ironand finally steel. It was in the period 1850 to 1890 thatwrought-iron reigned supreme. This paper focuses onmaterial properties and historical test data. The struc-tural uses of wrought-iron are described in compilationvolumes for the periods 1750–1850 (Sutherland 1997),and 1850–1900 (Thorne 2000) and also in previouswork by the second author (Swailes 2006, Swailes &Marsh 1998).

For most of the 19th century engineering quanti-ties such as yield stress and modulus of elasticity werenot measurable. Furthermore, quality of material var-ied considerably. The need for better understandingof the properties of iron became ever more impor-tant as engineers and architects designed structures ofgreater span and complexity. This prompted the spreadof experimentation in many countries.

2 COMPOSITION AND TEXTURE OFWROUGHT IRON

Wrought-iron is a composite material, not in thestructural sense but in the metallurgical sense. Itis composed of two phases, one being ferrite-ironwith a BCC crystal structure and the other slag.

Wrought-iron is relatively malleable and ductile dueto low carbon content, typically less than 0.05%. Steelhas carbon content in the range 0.2–1.0%.The slag isnot well bonded to the ferrite and so does not enhancethe strength of the iron. “The amount of slag in wroughtiron can be up to 3 wt% of the total. It is a glassysubstance composed of iron silicate and iron oxide”(Walker 2002).The thickness of the slag inclusions canrange from microscopic size to 3 mm. They appear asnarrow elongated strands or streaks and are given thisshape by rolling the iron in a particular direction whilethe iron is hot. By dividing the metal into strands of fer-rite the iron can be described as having a macroscopicgrain due to its fibrous appearance. This texture is bestseen when a nicked bar is bent backwards tearing openthe metal (Figure 1).

3 EFFECT OF SLAG INCLUSIONS ONDUCTILITY AND STRENGTH

Unlike steel, which from the 1860’s onward could beproduced in large quantities, wrought-iron was alwaysmade in small separate batches in the puddling furnace.In the puddling process, pig iron was converted intowrought-iron by removing carbon.The manufacture ofwrought-iron was described by Skelton (1924) and invarious works by Gale (1963, 1964, 1969, 1977). Themost obvious difference between wrought-iron and

207

Figure 1. Fibrous texture of wrought-iron revealed bytearing open a bar in a nick-bend test. (Thorneycroft 1850).

steel is that wrought-iron always contained slag whilesteel did not. Iron had to be hot worked or wrought inorder to produce a metal of better quality. Repeatedworking caused the slag inclusions to be refined. Theywere made smaller, shorter and more evenly dispersedwith a consequent improvement in mechanical prop-erties. It was shown by experiment that the benefitsof repeated workings reached a peak at about the6th working and thereafter the strength diminishedwith further reworking (Turner 1908). However, it wasnot economical to repeat the process beyond the 4thworking. (Gale 1964).

The size, shape and distribution of the slag inclu-sions in wrought-iron vary greatly. In cases wherethere is a high degree of elongation, the ferrite matrixis divided into columns. (Gordon 1988). In this con-dition low ductility can be observed in test sampleseven though the ferrite itself is quite ductile. Long fer-rite columns make the metal more prone to internalrupture. (Gordon 1988).

Excessively long slag inclusions, from insufficientworking of the metal, cause low ductility. Equally badare tiny, globular slag inclusions which result fromexcessive working of wrought-iron. An iron with sucha microstructure “lacks the fibrous texture, typicalof good wrought-iron, and behaves like a dirty low-carbon steel; it tends to be brittle and has poor fatigueproperties.” (Jeffrey 1959). That is why wrought-iron exhibits improved mechanical properties up toabout the 6th working and thereafter the mechanical

properties deteriorate with further working. Exces-sive working of iron may have occurred when rollingcomplex shapes.

4 COMPOSITION AND ITS EFFECT ONMECHANICAL PROPERTIES

The condition of red-shortness is a lack of cohesionwhen the metal is red-hot. The metal cracks or crum-bles when being hot worked. It is caused by an excess ofsulphur and results from the iron not being sufficientlypurified in the puddling furnace (Skelton 1924). Sul-phur is present in wrought-iron as iron sulphide (FeS)but it tends to segregate from the ferrite at the grainboundaries. Because iron sulphide has a low meltingpoint it causes a lack of cohesion between the grainswhen the iron is heated to red-hot (Johnson 1939).Sections rolled from red-short iron are likely to haverough edges (Johnson 1939). Cold short iron is thecondition of low ductility at normal temperatures dueto an excess of phosphorus or copper.

“A good wrought-iron would have a maximum sul-phur content of 0.05% and a maximum phosphoruscontent of 0.16%. The manganese content should beless than 0.1% and silicon content less than 0.2%”(Jeffrey 1959). Of all the impurity elements phos-phorus has the most significant effect on mechanicalproperties. Elevated phosphorus content causes higheryield strength and ultimate strength but causes a sharpfall in ductility and impact resistance (Figure 2).

5 THE EFFECTS OF COLD WORK

It is important that the metal is kept hot particularlyduring the final stages of working into a finishedshape so as to avoid strain hardening the metal by coldrolling. This can sometimes be difficult, particularlywith the rolling of long bars of small diameter, becausethe smaller or thinner the section the faster it loosesheat.

Annealing removes the effects of cold work. In anumber of the testing programs discussed in this paperthe samples were annealed before tensile testing. Thiswas done because the mechanical history of someof the specimens was unknown. By first annealingthe metal the ‘natural’ strength and ductility could bedetermined as opposed to measuring the strength of astrain hardened sample.

Shearing and punching iron strain hardens the areaaround the cut or hole. “M. Barba showed that cut-ting out a ring 1/8 inch thick round a punched hole, orannealing the plate, entirely removed the prejudicialeffect of punching” (Unwin 1910). Drilling holes forrivets did not strain harden the metal but was slowerand more expensive than punching. Kirkaldy found

208

Figure 2. Effect of phosphorus on Izod impact energy(Jeffrey 1959).

that punched plates experienced a 50% loss in ductil-ity compared with plates that were drilled. (Kirkaldy1876).

6 STRUCTURAL DESIGN DURING THE 19THCENTURY

The accuracy of testing machines prior to the 1860’swas generally not sufficient to determine preciselyelastic limit. At that time engineers based their designson the quantity that they could measure with certainty,which was the ultimate strength or breaking strengthof the material. In other words, in much the same wayas modern day engineers apply a factor of safety to thecharacteristic yield strength to obtain a working or safedesign strength, 19th century engineers applied a fac-tor of safety to the ultimate strength in order to obtain asafe design strength. Large factors of safety were usedto ensure elastic behaviour of the metal under workingloads.

Regarding the “question of safe working strengthmuch difference of opinion exists among engineers, thepermanent supporting power of iron being variouslyestimated at from four-tenths down to one -tenth of itsbreaking strength” (Colburn 1863).

In 1859 the Board of Trade in Britain imposeda value of 5 ton/in2 (77 N/mm2) as the maximumdesign stress for members in railway bridges which

became the standard value for all structural work.(Colburn 1863). French engineers worked to a value of3.8 ton/in2 (59 N/mm2) as the maximum design stress(The Engineer 1863).

These figures were based on tests that did notdirectly measure elastic limit. For example, whenDavid Kirkaldy conducted his investigation into theproperties or wrought-iron and steel in 1858–61 atthe Napier shipbuilding firm, his instruments couldnot measure elastic limit. When Napier and Sons dis-continued Kirkaldy’s testing program he resigned hisposition with the firm so that he could devote his entiretime to materials testing. He designed a new form oftesting machine and set up Britain’s first commercialtesting works in 1865 (Smith 1980). Records of histests conducted in 1866 show that by this time Kirkaldywas able to measure elastic limit.

7 THE UNITED STATES BOARD APPOINTEDTO TEST IRON, STEEL AND OTHER METALS

In America the desire for better understanding of themechanical properties of iron was probably greaterthan anywhere else, because at that time manyAmerican engineers viewed themselves as being tooreliant on foreign experimental work for knowledgeabout the mechanical properties of iron and steel(Pugsley 1944). The U.S. Government created a Boardin 1874 to provide a national facility for testing mate-rials. (Gordon 1996). Congress allocated $75,000 tothe Board and in June 1875 a contract was madewith Albert H. Emery to design a precision testingmachine (Gibbons 1934). The machine was com-pleted and installed at Watertown Arsenal, Watertown,Massachusetts in 1879 (The Engineer 1888).An exten-sive program of testing began which was reported on anannual basis. The testing machine at Watertown Arse-nal, which became known as the ‘United StatesTestingMachine’, was a significant achievement as it was oneof the largest and most precise testing machines in theworld. America was now in a position to make signif-icant contributions to the field of materials testing.

8 MATERIALS TESTING AT M.I.T

As in Britain and Europe the American universitiesalso set up their own materials testing laboratories. Onein particular was that at the Massachusetts Institute ofTechnology. This particular university is of interest inthe present context because of the meticulous recordsof tests on wrought-iron that were made there in the1880’s and 90’s.

The mechanical engineering laboratory at M.I.Twas established in 1883 by Gaetano Lanza. “Lanza wasborn in Boston in 1848. His father was an Italian count

209

Figure 3. United States Testing Machine at WatertownArsenal (The Engineer 1890).

who went to America as a teacher of languages. Hismother was fromVermont. Gaetano Lanza received hiseducation at the University of Virginia where his fatherheld the chair of professor of languages. Following hisgraduation in 1870 Gaetano Lanza became an assistantinstructor of mathematics at the University of Virginiabut resigned in 1872 to take up a position as instructorat M.I.T. He taught mathematics for a short time, butsoon began to teach mechanics, becoming professor oftheoretical and applied mechanics in 1875” (The Tech1925).

In 1883 Lanza was put in charge of the departmentof mechanical engineering and immediately began aconsiderable expansion of the mechanical engineer-ing laboratories. He was particularly interested in thetesting of full size structural members (Lanza 1912).At M.I.T. Lanza had a Fairbanks testing machine of50,000 lbs capacity which besides being used for mak-ing tensile tests on iron and wire rope, could be adaptedin such a way as to enable full size beams to be testedfor transverse strength and deflection. The allowablespans were up to 25 feet. (Lanza 1887)

Lanza retired from M.I.T in 1911 but for a numberof years after this he was associated with the BaldwinLocomotive works in Philadelphia, where his expertisein full size mechanical testing was required (The Tech1925). M.I.T.’s test records are now being used in thepresent research on wrought-iron.

9 TENSILE STRENGTH OF BAR IRON

The tensile strength of wrought-iron is greater alongthe direction of rolling than across it. This is because

Figure 4. Stress-strain graph of 7 round bars with diameters10 mm, 13 mm, 17 mm, 23 mm, 26 mm, 39 mm, and 50 mm(Watertown Arsenal 1888).

when loaded across the grain the slag filaments runperpendicular to the load path, and in this directionthey act as voids for the propagation of internal cracksacross the specimen. If loaded along the grain the fer-rite is more continuous with reduced tendency for theformation of internal rupture surfaces.

The tensile strength is greater in narrow bars andthin plates than in thick bars or large forgings. Thisis shown in Figure 4 where a series of bars of thesame material but with diameters ranging from 50 mmdown to 10 mm were tested in tension.The stress-straingraphs are staggered to show more clearly the reduc-tion in yield strength with increasing diameter. Thethinner bars have greater strength because they expe-rienced a greater amount of hot rolling which makesthe ferrite grain sizes smaller and also causes greatercohesion between grains (Johnson 1939).

Because different structural sections undergo dif-ferent degrees of working it is reasonable to attemptto identify certain ranges of strength for the differ-ent principal structural sections, namely bars, plates,angles and beams. Figures 5 and 6 and Table 1 give theresults of tensile tests on bar iron. In order to identifypossible regional variations the results were dividedaccording to country of manufacture.

From this data the Scandinavian iron has a lowerelastic limit but higher ductility than the British

210

Figure 5. Elastic limit and ductility ofAmerican, British andScandinavian bar iron. 390 test results. (O’Sullivan 2007).

Figure 6. Elastic limit ofAmerican and British bar iron. 355test results. (O’Sullivan 2007).

or American irons. Scandinavian wrought-iron wasprized for its high quality being tough and eas-ily worked. This high ductility was due to theScandinavian ores being naturally low in phosphorusbut also due to the use of charcoal rather than coke inthe smelting furnace (Fairbairn 1864).

Table 1. Summary of tensile test data on British andAmerican bar iron. Elastic limit values are represented inFigure 6.

Bar iron. Tested parallel to grain. Number of tests: 355

Elastic Ultimate Elongationlimit strength at failureN/mm2 N/mm2 %

Range 127–304 278–533 3.7–40.5Mean 204 353 22Standard deviation 35 26 8

Figure 7. Cross-piling to form plate-iron. (Hutchinson1879).

Toughness is the property that should be used inassessing the quality of wrought-iron. Toughness isdetermined by both strength and ductility, and eventhough the Scandinavian irons have the lowest strengththey generally have the highest ductility, and so,can be considered to be of good quality. However,if attempting to assign a characteristic strength toBritish or American wrought-iron it is reasonable toexclude Scandinavian iron from the statistical analy-sis. Figure 6 is a histogram of the measured elasticlimits of British and American bar iron only. The datarefers to tests along the grain direction. For compari-son the yield strength of chemically pure iron has beenincluded. This is iron with very low impurity contentand no slag.

10 TENSILE STRENGTH OF PLATE IRON

For plate iron, an effective means of equalising thestrength parallel and perpendicular to the direction ofthe grain was cross-piling, in which the bars were piledin alternating directions as shown in Figure 7 beforebeing rolled into a thin plate.

However, sometimes the plates were formed withthe outer layers in the same direction resulting ingreater strength in that direction (Figure 7). Further-more the direction of final rolling may have given somedominance to the grain in that direction. In the resultsthat follow the grain direction of the plate refers tothe dominant grain direction. It can be seen from thecollection of about 550 tensile tests results along the

211

Table 2. Summary of test data represented in Figures 8and 9.

Plate iron. Tested parallel to grain. Number of tests: 550

Yield Ultimate Elongationstrength strength at failureN/mm2 N/mm2 %

Range 160–363 232–470 1–36Mean 240 345 15Standard deviation 32 35 7

Table 3. Summary of test data for plate iron tested per-pendicular to grain. Yield strength values are represented inFigure 9.

Plate iron. Tested perpendicular to grain. No. of tests: 115

Yield Ultimate Elongationstrength strength at failureN/mm2 N/mm2 %

Range 154–298 183–389 0.1–29.2Mean 208 296 8Standard deviation 36 39 7

grain (Table 2) and about 115 tests perpendicular tothe grain (Table 3) that plate iron is on average about15% stronger in the direction of the dominant grain. Itis because of this that plate girders were constructedwith the grain of the plate along the longitudinal direc-tion of the girder. From the data the characteristic yieldstrength along the grain is 185 N/mm2 which is lowerthan the value of 220 N/mm2 given in the HighwayStandard BD21.

11 STRENGTH OF ANGLE, TEE, AND BEAMIRON

The tensile strength of angle and tee iron is greaterthan that of plate iron, which is expected, as it is rolledfrom piled iron layers of parallel grain. With regard tothe more complex rolled I-section too few data exist tomake any generalisation on this form. Tests conductedat M.I.T. on samples cut from an I-beam gave a meanyield strength of 165 N/mm2 while tests conducted atUMIST on samples cut form a Belgian I-Beam gave amean yield strength of 319 N/mm2 (Kontos 1996).

12 COMPRESSIVE STRENGTH OF WROUGHTIRON

Tests by Marshall in 1887 and Kirkaldy in 1866 showthat for practical purposes the tensile and compressivestrengths of wrought-iron can be taken as the same.However, Gordon has proposed that in cases where

Figure 8. Yield and Ultimate strength of plate iron testedalong grain direction. Vertical pairs of data points are fromthe same tensile test. (O’Sullivan 2007).

Figure 9. Yield strength of plate iron tested along and acrossgrain direction. (O’Sullivan 2007).

212

Table 4. Summary of tensile test data on angle and tee iron.(O’Sullivan 2007).

Angle and tee iron. Tested parallel to grain. No. of tests: 94

Yield Ultimate Elongationstrength strength at failureN/mm2 N/mm2 %

Range 193–351 301–448 4–37Mean 245 370 22Standard deviation 27 30 7

Table 5. Numerical summary of modulus of elasticity datarepresented in Figure 10. (O’Sullivan 2007).

Samples tested parallel to grain. Number of tests: 242

Modulus of elasticity kN/mm2

Range 124–253Mean 197Standard deviation 13

the slag inclusions are excessively elongated test sam-ples can exhibit lower compressive strengths due to theferrite matrix being divided into columns which canbuckle prior to yielding of the ferrite (Gordon 1988).

13 BRITTLE FAILURE OF WROUGHT IRON

Past failures of wrought-iron structural elementshave included suspension bridge chain links. WilliamKirkaldy investigated the failure of a tie-bar fromCharing Cross Station roof in 1905 (Swailes 2005).More recently a beam failure in a building in Leedswas reported (Bland 1984).

The consequences of failure are more severe forsome structures than others and some structures aremore important. To take account of this in assessmentdifferent factors of safety could be applied to differenttypes of structure. However, for reasons of simplicityand lack of sufficient data, use of a single factor ofsafety is prevalent.

Fatigue failure of wrought-iron is an area thatdeserves further investigation. Cullimore conducteduseful work on fatigue and determined that the fatiguelimit of wrought iron can be taken to be about one-thirdof the ultimate tensile strength (Cullimore 1967).

Lack of toughness rather than strength has beenattributed to various failures of structural elements.Wrought iron from the S.S. Great Britain (Morgan1996) and Walnut Street Bridge in the U.S. (Green1999) showed a high ductile-to-brittle transition tem-perature indicating that wrought-iron is potentially

Figure 10. Modulus of elasticity of bar wrought-iron(O’Sullivan 2007).

prone to brittle fracture at normal temperatures. Impacttest data indicates that the toughness of wrought isquite variable. Charpy values for wrought-iron froman American truss bridge were in the range 34–144Joules (Sparks 1998) while Charpy values for mate-rial from another American bridge were in the range10–80 Joules. (Green 1999). For a rolled wrought-ironbeam tested at UMIST the Charpy values were quitelow, 10 Joules for the flanges and 23 Joules for the web(Steude 2000).

14 MODULUS OF ELASTICITY

Tensile tests on various American, British and Nor-wegian wrought-irons were compiled to produce thehistogram of values for modulus of elasticity shownin Figure 10. The mean value is 197 kN/mm2 which isclose to the BD21 value of 200 kN/mm2. Outlying val-ues are a consequence of experimental measurementerror.

15 CONCLUSIONS

The quality of wrought-iron is highly dependent onmanufacturing practices. Firstly the skill of the pud-dler dictated composition while the care exercisedin the forge in piling and rolling iron determinedthe microstructure of the metal. Wrought-iron was abatch produced metal, which resulted in considerablevariation in mechanical properties.

213

Of the various impurity elements carbon and phos-phorus exhibit the greatest influences by causingembrittlement if present in excess quantity. Tests haveshown excess phosphorus to result in significant lossof impact resistance.

High tensile strength should not be used as a mea-sure of quality. Ductility must also be taken in accountso that together with strength it is toughness that shouldbe used as the measure of quality.

The UK Highway Standard BD21 quotes 220 N/mm2

as the characteristic yield strength of wrought-iron.The tests examined in the present research so far indi-cate the characteristic yield strength of plate iron to be185 N/mm2, and characteristic elastic limit of bar ironto be 154 N/mm2.

ACKNOWLEDGEMENTS

The authors are grateful to Cass Hayward & Partnersand to Network Rail for sharing the findings of a studyof wrought-iron stresses.

The Watertown Arsenal and M.I.T. test reports werelocated by Institution of Civil Engineers ArchivistCarol Morgan.

REFERENCES

Barlow, P. 1867. A treatise on the strength of materials.London: Lockwood and Co.

Bland, R. 1984. Brittle failure of wrought-iron beams. TheStructural Engineer. Vol.62A. No.10. October 1984. Let-ter to Verulam. Page 327.

Colburn, Z. 1863. On the relation between the safe load andthe ultimate strength of iron. The Engineer. March 6th1863. Pages.136–137.

Cullimore, M.S.G. 1988. Fatigue and fracture investigationcarried out on the Clifton Suspension Bridge’, Proc. ICE.Part 1. 84. (April 1988) pages 309–329.

Cullimore, M.S.G. 1967.The fatigue strength of wrought ironafter weathering in service. The Structural Engineer. Vol.45. No.5. pages 193–199

Department of Transport. 2001. The assessment of highwaybridges and structures, Departmental standard BD 21/01.

Fairbairn, W. 1864. Iron its history, properties, & processesof manufacture. Edinburgh: Adam and Charles Black.

Gale, W.K.V. 1977. Historic Industrial Scenes Iron and Steel.Buxton: Moorland Publishing Company.

Gale, W.K.V. 1969 Iron and Steel. London: Longmans, Greenand Co. Ltd.

Gale, W.K.V. 1963. Wrought iron: A Valediction. Transac-tions of the Newcomen Society. Vol.36. pages 1–11.

Gale, W.K.V. 1964. The rolling of iron. Transactions of theNewcomen Society. Vol.37. Pages 35–46.

Gibbons, C.H. 1934. History of testing machines for materi-als. Transactions of the Newcomen Society, Vol.15, Pages169–184.

Gordon, R.B. 1988. Strength and Structure of Wrought Iron’,Archeomaterials Vol. 2, No. 2, Pages 109–137.

Gordon, R.B. 1996. American Iron 1607–1900. Baltimore:The Johns Hopkins University press.

Green, P.S., Connor, R.J., Higgins, C. 1999. Rehabilitationof a nineteenth century cast and wrought iron bridge.Proceedings of the 1999 Structures Congress. Pages259–262.

Higgins, J.H. 1930. Investigation of the effect of phospho-rus in wrought iron. Contained in Appendix II of thereport of committee A-2 on wrought iron, Proceedingsof the annual meeting of the American Society for TestingMaterials, Part 1, June 23-27, 1930, Pages 159–186.

Hutchinson, E. 1879. Girder-making and the practice ofbridge building in wrought-iron. London: E. & F.N. Spon.

Johnson, J.B. 1939. Johnson’s materials of construction’Rewritten and revised by M.O. Withey & J. Aston, 8thEdition, New York: John Wiley & Sons Inc.

Jeffrey, R., Seager, J. & Woodhead, J.H. 1959. The mechani-cal properties of wrought iron. Safety in Mines ResearchEstablishment, Research report No. 154, Ministry ofPower.

Kirkaldy, D. 1876. Results of an experimental inquiry into therelative properties of wrought-iron plates manufactured atEssen, Rhenish Prussia, andYorkshire, England. London:Kirkaldy Testing and Experimenting Works.

Kirkaldy, W. G. 1891. Illustrations of David Kirkaldy’s systemof mechanical testing. London: Sampson Low. (located inICE archive)

Kontos, N. 1996. Investigation of wrought iron as a structuralmaterial. MSc Dissertation. UMIST.

Lanza, G. 1887. Notes on the Engineering Laboratories of theMassachusetts Institute ofTechnology. Minutes of the Pro-ceedings of the Institution of Civil Engineers, Part 1. Vol.91. Session 1887-1888. Pages 347–365. Paper no. 2245.

Lanza, G. 1912. The progress in testing full-size pieces underpractical conditions, together with locomotive testing inthe United States. Journal of the Franklin Institute. VolCLXXIV. No.6. December 1912. Pages 607–620.

Marshall, C.A. 1887. Compressive strength of steel andiron. Transactions of the American Society of Civil Engi-neers, Vol. XVII, paper no.364, August 1887, New York:American Society of Civil Engineers. Pages 53–110.

Morgan, J.E. 1996. The wrought-iron of the S.S. GreatBritain. Royal Institute of Naval Architects, confer-ence proceedings, Historic Ships Design, Restoration &Maintenance.

M.I.T. 1897. Tests of engineering materials made at theMassachusetts Institute of Technology, 1887–1897, Com-piled from Technology Quarterly, Society of Arts, Mas-sachusetts Institute of Technology. (Produced by M.I.T. inbook form. A copy is located in the ICE archives).

O’Sullivan, M. 2007. First year PhD report. University ofManchester.

Pugsley, A.G. 1944. The history of structural testing. TheStructural Engineer, Dec 1944. Pages 492–505.

Rawdon, H.S. & Epstein, S. 1924. The nick-bend test forwrought iron.Technological papers of the Bureau of Stan-dards No. 252, part of Vol.18. Washington: Dept. ofCommerce, Government printing office.

Skelton, H.J. 1924. Economics of iron and steel. 2nd Edition.London.

Smith, D. 1980. David Kirkaldy (1820–1997) and Engi-neering Materials testing. Transactions of the NewcomenSociety, Vol.52. Pages 49–65.

Sparks, S.P. & Badoux, M.E. 1998. Non-destructive eval-uation of a historic wrought-iron truss bridge in NewBraunfels,Texas.APT Bulletin.Vol. 24, No.1. Pages 5–10.

214

Steude, T. (2000), ‘The strength of wrought iron and earlymild steel beams’, MSc Dissertation, UMIST.

Sutherland, R.J.M. (ed). 1997. Structural Iron, 1750-1850.Aldershot: Ashgate Publishing ltd.

Swailes, T. & Marsh, J.O. 1998. Structural appraisal of iron-framed textile mills. Institution of Civil Engineers Designand Practice Guide. London: Thomas Telford Ltd.

Swailes, T. & Marsh, J. 2005. Development of long-span ironroof structures in Britain. Proceedings of the Institutionof Civil Engineers, Structures and Buildings 158. October2005. Issue SB5. Paper No.13908. Pages 321–339.

Swailes, T. 2006.Scottish Iron Structures. Guide for practi-tioners. Edinburgh: Historic Scotland.

The Engineer. 1863. The elastic limit of iron. The Engineer,March 6th 1863, Pages 139–140.

The Engineer. 1888. The Watertown Arsenal TestingMachine. The Engineer, January 20th 1888, Page 54–55.

The Tech. 1925. Professor-Emeritus Gaetano Lanza. TheTech. Monday. September 28th. 1925. Cambridge: Mas-sachusetts Institute of Technology.

TheTech. 1928. Obituary of Gaetano Lanza.TheTech. Friday.March 23rd. 1925. Vol. XLVIII. No.19. Cambridge: Mas-sachusetts Institute of Technology.

Thorne, R. (ed). 2000. Structural iron and steel 1850–1900.Aldershot: Ashgate Publishing Ltd.

Thorneycroft, G.B. 1850. On the manufacture of malleableiron; with the results of experiments on the strengthof railway axles. Proceedings of the Institution of CivilEngineers. Vol.9. Pages 294–300, Paper No. 828.

Turner, T. 1908. The Metallurgy of Iron. 3rd Edition, CharlesGriffith & Company Ltd. Strand.

Unwin, W.C. 1910. The testing of materials of construction.3rd Edition. London: Longmans, Green & Co.

Walker, R. 2002. The production, microstructure and proper-ties of wrought iron. Journal of Chemical Education, Vol.79, No. 4, April 2002.

Watertown Arsenal. 1883-1893. Report of tests on thestrength of structural material made at the WatertownArsenal, Massachusetts. (Reports appeared as annualvolumes).

215