The following miscellany is intended to report progress in studies into the background of Wilkinson's iron boats, since the paper in Wilkinson Society Journal No.15, 1987. The first item especially is I believe of some considerable relevance to the Trial itself, both in terms of metallurgy and of specific construction details. Strict proof of a direct relationship is of course another matter.

As no formal investigation of the cast iron water tanks at The Lawns (the "soft water furnaces" of the 1800 inventory - Wilkinson Society Journal No.7, 1979, page 5) seems likely at present, I made some rough measurements, supplemented by photographs, as a record, and offer these brief notes on what proved to be intriguing artefacts. Mr Michael Berthould kindly assisted; he had already established the valuable information that the smaller roof tank bears the cast maker's name - Bancks & Co. We may hope that he will shortly publish his research into the background of these tanks, as revealed by this nameplate.

Two small samples of loose material were very kindly examined by Mr Brian Gilmour, of the Royal Armouries, and his report is presented separately. [Appended below]. The results are of considerable interest.

The tanks are best described by the accompanying drawings, but the construction is essentially by bolted flanged construction. The smaller of the two tanks, that bearing Bancks' name, is conventional for its period. The larger tank, however, has a curious integral cast joggle in both side and bottom plates, giving flush sides and bottom externally, and is thus particularly interesting. It appears that similar joints occurred widely in the shroud plates, at least, of iron waterwheels - as at Bersham - and in many photographs. At Bersham they also appear in the sole-plates and bucket castings. There is in practice no real saving in weight over the use of flanges as elsewhere in the tank; and the detail looks rather weak. It might be viewed as most appropriate in a long modular structure such as a canal barge: it seems curiously out of place in the sides especially of a tank in which other plates are large enough to span the whole length of the side. It represents a curious experiment, and not one that could be readily made in an open sand mould ?

A 6 foot length of cast iron pipe of 2-1/4 inches outside diameter acts as a spacer for a tie bar across the top of the tank. It has a typical socket detail at one end, and two feeder sprues at mid-length. This tank rests upon three lengths of heavy cast iron channel section, 6-1/4 x 2-1/4 inches.

The dimensions of this tank, too, are curiously suggestive of the dimensions of the Trial: 76 inches overall width and 40 inches deep. Perhaps the wooden lining of the gunwales was actually external in the Trial, rather than internal, as I had supposed in my work for the Wilkinson Society Journal No.15 ? Actually, this would serve two very necessary functions in this case. First, to protect the boltheads, which being sharp-edged and protruding half an inch (at least in this tank) would be very vulnerable to being sheared off against quays and lock walls; and secondly, to provide a softer surface to protect the iron from the worst of impacts against structures and other boats. The timber required to fit the dimensions of the Trial (with five-sixteenth inch plates from the same pattern) would be 2-1/2 inches, which is a quite reasonable result. A recalculation of weights and capacities of the Trial would be necessary: the assumptions here differ from those used in the Wilkinson Society Journal No.15.

The one factor that contradicts any direct relationship with the Trial is the plate thickness - nominally half an inch, though calipered between 3/8 and 5/8 inches. This average thickness (almost twice that recorded in the Birmingham newspaper account) would lead to an impossibly great weight in the Trial. The large plates exhibit surface irregularities characteristic of open sand moulding, and the thickness could presumably have been varied to suit the purpose, while retaining the same pattern.

One of the samples analysed was from a broken fragment of a plate, found lying in the debris in the tank. It is reported as grey iron, but with a carbon content of only about 2.5%. Consequently it is virtually proven that the metal used was remelted in a reverbatory furnace, and certainly not cast direct from a blast furnace. It thus provides the first definitive evidence that "softer" cast iron (a relative term, of course, as the local fracture damage shows) was being used for thin plates of dimensions appropriate to barges, at about the right date, and probably in the right area. A full comparative study of the metallurgy of both tanks would be of considerable interest. Were both tanks by Bancks; were Wilkinson and Bancks connected; can the tanks be more closely dated; were they erected in the same fashion; have they been moved ? Certainly there is much interesting detail concerned with pipe connections currently obscured by their location.

The joggle detail would look relatively stronger with a thinner plate. The joints are packed out with what appeared, superficially, to be wrought iron - certainly hard, iron-stained and partly laminar. The most likely explanation would then have been that this was intended to provide a malleable surface that could be beaten up to seal the gaps - as in later riveted steam boilers; or in a run-lead joint, still occasionally used in water mains. The bolts on the other hand, from their superficial corrosion pattern, and accurately square faces and arrisses, look suspiciously like cast iron. Malleable iron bolts are not unknown today.

However, the strip turns out to be a severely corroded cast iron, of slightly lower carbon content than would be expected from a blast furnace. This rules out any connection with sealing the joints, and focuses attention on the question of the patterns used to cast the joggled plates, in particular. Another method of sealing the cast iron plates of an aqueduct that has been noted was used by Outram on the Huddersfield Narrow Canal in 1800. He used "rust joints" between the flanges, consisting of iron turnings and borax made into a paste (R.B.Schofield, in Transactions of the Newcomen Society, Vol.53, pp 17-38).

Without disturbance to the tanks I can take these points no further, and many questions must await a formal study by specialists with laboratory and conservation facilities. The tanks are severely corroded internally, and there is some local damage at corners.

The third tank, at ground level, is in my view a much later construction, much on the lines of a Braithwaite tank, though very delicately detailed, with scalloped edges to the flanges, square cored bolt holes, and variations in thickness.

It has been mentioned to me by Mr Eric Cox that other cast iron water tanks have been noted in the Broseley area in the past. Any additional information would be welcome.

The discovery of these large thin joggles plates, of low carbon cast iron, taken with the existence of curved plates of even thinner section in the Bersham waterwheel, noted by Mr Douglas Braid, is strong supporting evidence for the practicability of construction (at least) of a cast iron barge in 1787. In particular, the joggled joint requires some explanation, and it is tempting to link its pattern directly to the Trial itself.

Robert Bell (or Bill) was a Londoner, and evidently a builder of small wooden vessels, from the content of his Patent. He was also, taken at face value, but bearing in mind that his Patent was approved ("inrolled") in June 1822, something of an innovator in the relatively new art of iron boatbuilding. If so, his case represents a little known area, covering the practical transition from wood to iron. The text, as published in the London Journal of Arts and Science, Volume IV, 1822, is interesting in its own right, and also poses questions about the construction of the Trial. For this reason it may be worth citing in full (as published it is largely an abstract):-

The subject of this patent is the construction of boats and barges with wooden bottoms and iron sides. It appears that three objects are proposed by the construction of iron vessels, viz, cheapness, lightness, and durability.

The mode of construction proposed in this specification applies to canal boats, which are stated as being generally seventy feet long, seven feet wide and four feet deep. The bottom is to be made according to any of the ordinary modes now in use. If the several planks are grooved and tongued, thin slips of iron are proposed to be used between each plank, instead of laths, which is the ordinary mode. Instead of raising the sides with planks, iron plates are employed, whose weight is about nine pounds to the superficial foot. These plates should not exceed two feet ten inches in width, and in length an allowance of three or four inches must be made for the rivets, so as to effectually secure the plates to the bottom.

Iron ribbed knees are then to be placed at such distances from each other as to receive the edges as to receive the edges of the plates upon the middle of the knees; holes, at proper distances, are to be made on each side of the middle rib of the knee, and corresponding holes in the plates: the whole is then to be riveted together in the usual manner, so that the vessel may be water-tight. It may be desirable to countersink the rivets into the plates, to preserve an even surface. The feet of the knees are to made of sufficient strength to allow of their being secured by rivets to the boat's bottom. The stem and stern post may be constructed in the usual way, or in any other manner that the purposes of the boat may require.

"I also claim, as new, the use of knees made of any iron, whether the rib be on the middle or the side, when they are applied for the purposes of connecting the plates which form the sides of the boat, barge, or vessel, with each other, and connecting or securing them to the bottom of the vessel, boat, or barge."

It is proposed to strengthen the sides by riveting along the upper edge or gunwale of the boat "angle irons", thin bars, or narrow plates of iron, turned over at right angles, the section which would be nearly represented by the letter L, which will preserve the form, and aid in resisting the blows to which boats are liable: this part, however, is not claimed as new. Similar "angle irons" may also be introduced diagonally between the knees, so as to add strength to the side of the boat, if that should be required.

In certain formed vessels, as in the London coal barges, the bottoms of the fore and after parts of the vessel are greatly exposed to the alternations of wet and dry, and these, when constructed of wood, such as elm, ash, beech, etc, will very soon decay. It is therefore proposed, in order to render the wood bottoms of such vessels more durable, to impregnate them with varnish, which may be effected in the following manner, and "which I believe to be new, and, if so, claim the invention."

"I boil the timber, or planks, in a mixture of pitch or tar, raising the heat to 350 degrees and upwards; this heat is continued from six to twelve or fourteen hours, according to the substance of the timber. Any other material which would bear the required heat would answer the purpose, but I prefer coal pitch and tar. I then rapidly draw off the hot stuff, and immediately cover the timber with thin varnish. I take care that little heat escapes during the change, for as the moisture and air contained in the timber are greatly expanded by the heat, so, as the heat decreases, the air and moisture will collapse, and leave room for the thin varnish to enter the timber. The heat is discontinued when the varnish is added, and the timber remains until it becomes cool. One part of coal tar, and five or six parts of spirit of coal tar make a good varnish, which I have used with full effect."

"The manufacture of such iron as I use for knees is entirely my invention; and I claim the application of it for the constructing of knees, the object of which is to increase the strength of a bar by the least possible additional weight where the bar is to sustain pressure by its flat bearing, and is thus effected: I take a bar of the proper size for the boat's knee, four inches broad and three eighths thick; another bar, called the rib, two inches broad, three eighths or four eighths thick. This rib is placed on the middle of the fore bar; they are then heated in a proper furnace to a welding heat, then passed through a pair of rolls with a proper groove in the lower one, and thus receiving sufficient pressure, the union of the two bars will be effected. Bar of any required size may be produced by the same process, by a proper adaptation of the machinery and the substance of the bars. But I find for the boats described before, if the face bar be from three to four inches broad and three eighths thick, the rib two inches broad and three eighths thick, the knees, when bent, will be sufficient."

The text then is not without interest for wooden boatbuilding too. The stricture about the formed ends of some barges was particularly applicable to Severn barges. These were broad and shallow, and as photographs prove even the planks adjacent the keel could curve up above the unladen waterline. These "garboards" in a more conventional hull were often of elm, to save the cost of oak, but this could not be done on a Severn barge. To compound the problem, Severn barges were often stranded high and dry for long periods, and only the most durable timbers would long survive such treatment, without even the benefit of salt water. In the older form, which was that used on the upper river to the end of the river trade, clinker strakes from near the bilge reached the stem well above the waterline. Bell's solution to preservation sounds a particularly unhealthy and dangerous process; but it was an economically significant area, and had attracted the attention of learned men for centuries, including the Royal Society.

The fact that "angle irons" need emphasis, and the lack of uniformity in nomenclature for the flange of the tee, variously fore and face, reflect the fact that 1822 is indeed an early date for rolled tees, though they are known to have been used in the funnel of the Aaron Manby early in 1822. Bell's angle irons are still "turned over", rather than rolled.

It seems that his plates were placed vertically, and so were at least four feet long. They were riveted directly to the tees to give a flush joint - whence the remark about countersinking the rivets. The availability of large rolled plates has made the use of iron competitive: not something that was true for John Wilkinson 35 years earlier.

It is not clear what the attraction of composite construction was: wholly iron construction was in use from at least 1804 in the Midlands, for similar vessels, and the triple advantages of weight, cost, and durability of iron vessels have been cited by Bell himself.

It is equally difficult to know what conclusions can be drawn from the granting of this Patent, as regards the construction of the Trial. It might be taken to rule out composite construction (if that were really in doubt), and also the provision of iron knees to support the side.

Almost more interesting, however, is the reflection that John Wilkinson did not patent his methods of barge construction. He had after all patented his earlier boring machinery, albeit to have it overturned in the national interest. Or turning that into a question: has anyone ever looked for a Patent for the Trial ?

It may be worth citing the examples of two early American iron vessels to supplement our understanding of the events and technologies surrounding the Trial - albeit of a later generation.

In 1825 a small iron steamer, the Codorus was built at York, Pennsylvania, by John Elgar, for use on the difficult Susquehanna River, of which the Codorus was a creek. She was the first iron vessel built in America, but had been preceded by some two hundred wooden steamers. Draught of water was again critical - five inches unladen is cited. She was roughly 60 x 9 x 3 feet - no precise dimensions survive. However, a crude sketch survives of her under construction: upside down, of riveted iron plates on angle ribs bent from flat strip, at about one foot centres. The purchase orders for the material indicate plate widths of about two feet: most of it was to be rolled to a gauge of 1/12 inch, the rest to 1/8 inch - and the gauge to be returned to the purchaser. The ribs required strips 7 feet long, 3 inches wide and 1/8 inch thick. All materials were closely specified.

Material for a boiler was also purchased, in 1/4 inch plate except the boiler head of 2 foot 10 inches diameter, 3/8 inches thick - if necessary formed from two plates trimmed with an overlap. Her total weight, complete with a 100 psi 8 HP engine was 4 tons.

She was, when complete except for her machinery and thus weighing approximately two tons, placed on two wagons coupled together, with timbers laid on them, and a bolster supporting a platform for the boat on each of them - on which the boat travelled 12 miles. Both the mode of transport and the extreme lightness are interesting points in relation to the Trial. Many further details, and numerous references to both American and British sources for early iron boats will be found in an article by A.C.Brown, The Sheet Iron Steamboat Codorus, in American Neptune, Vol X, 1950, pp 163-90.

A second account concerns the John Randolph, an iron steamer built at Lairds, Birkenhead of best British rolled boiler plate iron, dismantled, and reassembled by riveting in Savannah, Georgia, in 1834. She was built for the cotton traffic from Augusta, and again draught was critical - wooden vessels were stranded at low water periods in the rivers. This vessel was 110 feet x 22 x 8 deep in hold, with bottom and lower strakes of 5/16 inches and the rest 1/4 inch. The unladen draught recorded was 2.75 feet - "about half that of a wooden vessel of the same size". These great draughts are explained by the existence of extensive superstructure for passenger accommodation, and also by at least 17 tons of boiler and machinery. The hull weight must have approached 150 tons, and the cargo capacity was well over 200 tons.

A similar vessel of 1836 had the hull divided into three watertight compartments (a very early occurrence), and plates of about 24 x 18 inches, overlapped 2-2.5 inches, and heavy iron frames at one foot centres.

These details are taken from the Georgia Historical Quarterly, Vol XXXVI, No 1, March 1952: an article by A.C.Brown, The John Randolph, America's First Commercially Successful Iron Steamboat. Curiously, Lairds exhibited a model of this vessel at South Kensington in 1876 (described in the 1878 catalogue). This model was loaned to Liverpool in 1886, and was returned to Lairds by 1889, but by 1953 its whereabouts were unknown.

The Times of 21 April 1990 carried an article related to the bicentenary of the launching of Greathead's lifeboat for the Tyne in 1790, following a design competition in 1789. Adrian Osler has done research into the background to Wouldhave's proposals for that competition, and to the lifeboats subsequently developed by Greathead. It emerges that Wouldhave's proposal, for which a model was submitted, was for a metal boat. Osler is reported as regarding this as way ahead of its time - but we may wonder. The York boat of 1777 may well have inspired the idea: the lifeboat was only to carry 24 people, not so very much larger. The weight and nature of the commercial Trial really rules it out as a prototype for Wouldhave, but both boats were nationally publicised, and prior illustrations of the possibilities of using iron boats.

William Wouldhave (1751-1821) actually proposed to use copper; his model was of tinplate. There is negligible additional information in Osler's book as to metal in boatbuilding.

The principal point of interest in Svedenstierna's original Swedish passage about Wilkinson's iron boats at Bradley (Wilkinson Society Journal No.16, p5), was not another translation of the whole, but the precise word used for iron plates. In the event, "jernplåter" is not definitively wrought or cast iron, but in modern usage would tend to imply thin sheets rather than cast sections, as far as I can establish. However, the original Swedish contains no other new information about the Trial.

In: Wilkinson Studies, Vol.I, 1991, Merton Priory Press, pp63-65, including 3 illustrations.

Two samples of water tanks dated to pre-1800 from The Lawns, the former Broseley home of the iron founder John Wilkinson, were examined at the Royal Armouries. The tanks were made of plates of metal held together with thin additional pieces located between flanged edges. One sample each from one strip and one plate were taken to identify the metal which was thought, possibly, to be cast iron in each case. The sample from the strip (AM252) was found to be almost totally corroded although small islands of surviving metal were found in the groundmass of corrosion products (Fig.1). The structure of some of these particles of metal had partially broken down although others survived unaltered to show the strip to have once been a piece of grey cast iron. Under higher magnification (Fig.2) a typical grey cast iron structure can be seen with graphite flakes in a mainly pearlitic matrix. Between the darker pearlitic areas are some paler areas which appear to be filled by the ternary phosphide eutectic (Fe-Fe₃C-Fe₃P) steadite. The structure would appear to indicate a carbon content of about 3.0-3.5% although the phosphorous and silicon levels were not determined so that the carbon equivalent (CE) value may be higher. Etching of the polished sections was done with 2% nital.

The sample from one of the tank plates (AM253) was a complete contrast in that the metal was very well preserved. This too was a grey cast iron although the graphite flakes in the mainly pearlite matrix were finer and much less numerous (Fig.3) than for the strip. Small paler areas of what again appeared to be the ternary phosphite eutectic, steadite, were visible in the mainly grey pearlite matrix. The number and fineness of the graphite flakes would suggest a carbon content of about 2.0-2.5%, quite low for a cast iron. The phosphorous and silicon levels were again not determined so that the carbon equivalent (CE) value may be higher.

It seems curious that the metal survival of the two pieces should be so different but one possible explanation is that the strip examined has acted as a sacrificial anode in an electrolytic corrosion process. If we assume that the plates of the tank are all similarly low in graphite to the sample then they would tend to be more resistant to electrolytic corrosion. The much more plentiful graphite of the strips (assuming these are all similar) would promote galvanic corrosion within these pieces which in contact with the more corrosion resistant plates would tend to act as sacrifical anodes and, therefore, corrode much faster - in proportion of the surface area of the tank plates to the flange strips. The sample of flange strip was still quite dense and, held in place, may still have been fairly impermeable, even when extensively corroded, especially if the joins were caulked as well.