A paper first presented at the BAFRA 5th annual conference, "Putting it into practice", London Guildhall University, 10 March 2000, by Ivo van Sandick, student furniture restorer at West Dean College. Previously published in Splinters, the BAFRA Student Section Newsletter, issue May 2000, and in Conservation News, the UKIC Journal, issue November 2000.
Thank go the chairman for his introduction, leading to the observation I was surprised to learn the subject of straightening warped timber has never before been on the agenda of one of the previous BAFRA-conferences. Deformation of wood is a major problem in the conservation and restoration of wooden objects, if not the biggest problem of all. But on the other hand, a simple reason for this could be that the subject has been avoided for reasons of complexity of work and devided opinions touching on the ethical aspect. One thing is clear: if the wood wants to bend, it bends. The piece that we see here has bent because of the branches. It becomes clear that the reaction wood around knots shrinks more than the rest.
Before we go into the different methods of straightening, non-intrusive and more intrusive, let's ask ourselves: what is wood? It is a sponge like material made of cells and fibres that swell and shrink depending on the temperature and moisture in the surrounding air, although to different extents in different directions.
When a tree is felled, first of all the water in between the cells evaporates, then shrinkage sets in as some of the water within cells dries out. Drying from its green condition, wood will retain its original dimensions until the moisture content falls to about 30%, from which point it will shrink as it dries. The moisture content is expressed as a percentage of the dry weight of the wood: a piece which is half dry wood fibre and half moisture will therefore have a moisture content of 100%. A certain percentage of bound water (that is the water within the cells) always remains. With thorough air-seasoning, the moisture content of most types of wood will fall to between 23 and 17%, at which level it will be in balance with the moisture content of the air around it, and would, in those pre-central heating days, be stable enough to be made into furniture.
In your careers as restorers you may have noticed that much antique oak furniture has not suffered from warping as much as other timbers, whilst mahogany furniture for example seems to have more than its fair share of such problems. The principal reason for this is that much of the wood used for oak furniture was quatersawn, so that the width of the board followed the radius of the log along the medullary rays, at right angles to the annual rings. This cut produces a board which is still liable to shrink but not distort. This popular cut has a name: it is called wainscot after the Dutch word 'wagenschot' and another reason for its popularity is that the medullary rays show quite attractively. There are several explanations for this name; the most believable one is that a 'wagenschot' is the shaft that connects the horsespan of a chariot or 'wagen' with the moving axis between the front wheels. It is evident that this shaft had to be extremely strong. Therefore the wood was riven rather than sawn, producing a piece with the longest grain possible.
When a panel was made from such quartersawn or wainscot boards, it would be fitted in a rebate frame without glue so that with seasonal environmental changes it could move freely. If such a panel has split, it is probably because movement has been inhibited by accumulations of dust and grit.
In the case of mahogany furniture, it seems that there was a demand for the widest board possible, which often meant taking a piece from the full thickness of the log. Such a board was cut through the growth rings of a tree at varying angles across the width, therefore exposing it to problems of distortion. Maximum shrinkage takes place at right angles to the radius, so while the quartersawn board loses a bit of its thickness, a tangentially cut board will tend to warp. Sapwood dries faster and shrinks more than heartwood. Easy to remember, it is obvious that the annual rings tend to straighten in this process.
A cool environment with a relative humidity of about 55% and a little bit of air movement (draught) is said to be ideal for wood and antique objects in general. Your customers want to be informed about this. Also advise them to turn up their thermostats gradually as autumn approaches and lower the setting step-by-step when spring comes. A sudden change in the circumstances is the worst thing possible as the moisture will be sucked out of or pushed into the material far too quickly causing even more tension inside it.
Now that we have an idea why and how wood warps, it is time to do something about it. Step one is to look at the role of an object and decide whether any action is necessary. If so, step two is to investigate the possible ways of gaining access to the warped component whilst considering the consequences of any other components involved. Then, having decided on the kind of treatment, we must think of a way to stabilize the object in its new shape and how to re-assemble whatever had to be taken apart.
Non-intrusive treatment would be changing the environmental circumstances, with little risk of damaging the object further except perhaps for transport. When a piece of furniture with a lot of shrunken components is placed in a damp environment, it might just return to its original shape.
When changing the environment is not enough, we can either try to straighten the piece of wood or adjust its context accordingly. In the case of a twisted door which does not close properly anymore, shifting the hinges is a possibility. In the case of a warped drawer, we could consider doing something about the drawer runners.
Depending on the reason why the wood has twisted, we can extract more moisture on the convex side or add moisture on the concave side. Humidity and heat partially soften the lignin and hemicelluloses in the wood, allowing the wood fibres to slide along each other. After cooling and drying, they return to their initial state and the wood maintains its new shape. Forcing the wood into a slight countertwist in the cooling process is an aid but not a guarantee to prevent twisting back.
Extracting moisture out of a convex surface could be undertaken with a household iron. It takes precise judgement to warm up the wood evenly with this method. Other options involve a hot plate or food warmer where we can leave the wood for a certain period to warm up and dry out gradually, or an electric blanket. Wrapped in such a blanket we can leave the wood clamped in a certain postion.
Introducing moisture on the concave side can be done by laying a steaming towel or paper over the surface to be treated, and covering the whole with a plastic sheet to slow down the evaporation of water. When the panel to be treated is a thick one, sometimes the use of steam is required, for which purpose we can use a steam gun or a steam chamber. Cold water vapor is produced by vibrating the water up to 20000 Hertz and thus shattering the molecules. Cold steam does not penetrate the wood as much as normal steam.
Using the steam chamber method, the moisture content needs to reach 25% with a temperature of about 100 degrees Celsius and these conditions have to be maintained for twenty minutes per centimeter thickness to make the board flexible. During this process this timber is pressed somewhat beyond a flat into the opposite direction of the original cupping. Make sure that the press does not stop air ciculation.
After this, setting of the timber in its new position takes place by leaving it in the press for about a week with a temperature of about 65 degrees Celsius, then releasing the press to monitor the outcome. If the cup reapears, the press is retightened and the whole is left alone for another week. Finally, the pressed timber is left to dry gently for two or three weeks with a temperature of about 25 degrees Celsius. Then the pressure is removed. Depending on the given situation, it is recommended to either reassemble the piece of furniture immediately to prevent the possible comeback of the cupping or to wait a while and see whether the cupping reappears.
A very common problem is the warped tabletop, which has been polished on one side. Moisture does penetrate shellac and all the other classic finishes, but at a highly reduced speed. So the underside will have moved more than the top.
If concave cupping is extreme and the integrity of the structure is compromised, a treatment of repeated wetting and drying the convex surface may be applied. The concave side serves as a compression-inducing restraint because it does not change in dimension. This is called compression set shrinkage.
Generally speaking all these solutions are only temporarily, as the wood will want to return to its disfigured state sooner or later. This phenomenon is called spring-back. The chances of experiencing a spring-back reaction are reduced when we countertwist the wood and leave it countertwisted for a long period of time, say a week or two. Because not only the cells on the concave side are compressed, but those on the other side as well, the tension in the wood is spread more evenly and chances of a return of the problem are minimized.
The sides of cupboards are just as eager to warp as table tops. As they have been veneered and polished on one side only, and have been exposed to central heated conditions, their disfiguration can be quite severe. When the panel is flattened using heat and moisture, it may be a good idea to glue a counterveneer on the inside: first a layer is applied with its grain perpendicular to the grain of the board, then a second layer of veneer with the grain running along with the board will hide the first one. In this way the panel is very unlikely to bend back.
Sometimes intrusive mechanical intervension is necessary. When there are no alternatives such as adjusting hinges or using moisture other techniques will have to be applied. Routering grooves then filled with battens may be the way to go, or sawing the panel through the middle. That way we get two flexible panels which, after light sanding, can be glued back with a doubleknifecut veneer in between the two sides to compensate for the thickness of the blade. Usually a special glue such as cascomite is used for additional strength.
Let's look at the case of this longcase clock door. The maker has not chosen his wood very carefully as the door has developed a propellor shape: the first 10 cm from the bottom were straight with nasty twists following, finishing off with a cupping at the top. The door is made of oak with mahogany veneer on both sides. It has solid mahogany mouldings around the edge displaying a beautiful and rich patina.
As both surfaces, front and back, were not to be tampered with, the most ethical and aestheticly pleasing solution also became the most intrusive. The door was cut in half.
The restorer considered cutting the door from the top down to the point where it was flat. That way he would have a register and slipping of wet surfaces could be avoided. Problem: how should the glue and veneer be introduced? In smaller panels the idea of sawing down to a certain point and leaving the two halfs connected may have its advantages, but with this case it was decided to saw it through completely. A point fence was made out of plywood to allow the cut to be continously adjusted to take account of the warp as the saw progressed through the door. The restorer offered the door up to the saw with the back held hard against the fence, and an assistant inserted a wedge in the fresh cut to make sure the saw blade would not be squeezed between the two loose sections. It became immediately obvious just how much tension there was in the wood as the two sides sprung away from each other. A fact which had been hidden by the veneers was that the core had not been cut >on the quarter= but slightly off, and that one of the sides was next to a knot. So the cupping and twisting was easily explained.
It became clear fairly quickly that the two halfs were not flexible enough; grooves had to be cut 2,5 cm away from each other at an angle perpendicular to the direction of the warpage. This happened whilst the board was clamped down in a 4 mm countertwist. The depth of the grooves was one third of the thickness of the side, hopefully leaving enough wood beneath it to prevent the grooves showing through. The next stage was to make up the battens from well seasoned oak. Cut slightly shorter than the length of the grooves to avoid problems with further shrinkage, these were glued in and brought to level with the surface.
A ply of three sheets of quartersawn oak was glued using cascomite, so that the outer two sheets would lay longitudinally, still in the countertwist position. When the cramps were removed, the twist had gone! Finally, the second half was glued to the first and held in a flat position for 24 hours. After colouring the sides the result was dashing.
Think of a situation where we cannot use mechanical methods of straightening. In such a situation, some magic may be done using polyethylene glycol (PEG). PEG is a watery polymer that has been developed by the Frenchman Charles Wurtz in 1859, but came into use for conservation of particularly waterlogged timber in the second half of the 20th century. In theory it is reversible, but once it has penetrated the wood, chances are it will stay there. Imagine trying to dissolve and remove an impregnated material from frail timber with veneer and a historic surface finish... Reversibility sounds great, but in practice the removal of PEG from a piece of wood would probably cause worse damage than before as it would need the introduction of a lot of water. So, given that practical reversibility does not exist and that it is imperative to consolidate rather than loose an object, we might as well be a little bit more intrusive. PEG differs from other common wood consolidants in its strong chemical affinity for cellulose. It swells the wood cell walls where as other stabilizers simply fill the voids between the cells. Softwoods tend to respond better than hardwoods, perhaps because there is relatively more cellulose in it. The same goes for reaction wood, where the cells walls are the thickest.
H H H H | | | | H - O - C - ( C - O - C )n - C - O - H | | | | H H H H
There are several types of PEG available, depending on the n in the chemical formula HO-CH2-(CH2-O-CH2) n-CH2-OH. The molecular weight may vary from 200 up to 58000. The most common types used in furniture conservation are PEG 400, which is liquid at room temperature, PEG 1000 which comes in blocks or flakes, and PEG 1500 in the form of pearls. In fresh or only slightly degraded wood with swollen but still intact cell walls, low molecular weight PEG can partly replace the water within the cell walls and thus keep them swollen when the residual water evaporates. In case the cell walls have lost their mechanical strength as in waterlogged timber, a high molecular weight PEG is needed to bulk the cell lumina.
The lower the molecular weight, the more effective the treatment is, but a disadvantage of low MW PEG is that it is also more hygroscopic. Wood treated with PEG 400 could therefore become damp in humid conditions perhaps too easily. A great deal of research is still needed in order to refine a warpage treatment from a trial and error process into a predictable and reliable procedure. Many variables affect the outcome: species, density, and thickness of the wood, the glycol MW, temperature and concentration of the solution, the effect of pre-swelling with mere water, etcetera, but what is known now is promising.
How do we use PEG? First of all, the concave surface to be treated is wetted thoroughly using towels or paper and covering the whole with a plastic sheet to slow evaporation. Where there is a glue joint or something like a veneered edgebanding, a parafin dam can prevent the moisture from wetting the veneer and glue. Also knots and burrs are to be avoided, as they will cause severe problems when the cells swell and the grain straightens, as you can imagine. After several hours, a reduction of the warpage can be seen.
Meanwhile, prepare a 30% solution of PEG. It dissolves easier in warm water, but should not be heated beyond 50 degrees Celsius. Gloves should be worn and contact with metal avoided as this will oxydize the PEG. Next, the solution is applied onto a clean tissue covering the wood and the plastic sheet is put back on top. Add some more the next morning, at noon and in the evening for the next four days. And be sure to cover the whole again with plastic after each application. Then, increase the concentration of PEG by applying it with a brush directly on the wood, and repeat this again several times a day for the next couple of days. Now a cellophane sheet is replaces the plastic sheet to allow some slow evaporation to happen. Once a blanched appearance has been obtained, the piece is left to dry slowly under the cellophane. Finally, the excess is removed with a damp cloth.
A few small remarks have to be taken into account: the treatment should be homogeneous and superficial, for if the penetration is too deep or unevenly spread over the wood, the treatment will lead to even more warping and other catastrophic results. It has been noticed that glycol continues to slightly swell a treated board for several months after treatment, presumably because there is free glycol in the cell voids which continues to be taken up into cell walls. On the other hand, it has also been noticed that the effect of a treatment using very low molecular weight PEG is reduced in time. Panels with PEG below 300 MW cupped back and this could be due to its volatility. A sealing agent may reduce this reduction.
As the PEG has a slightly greasy property, wood treated with it will not accept PVA glue but it will accept animal glue, although to a slightly lesser extent than untreated wood. Reglueing of a veneer on a panel treated with PEG may have a negative effect as the PEG is not able to counter the forces exerted by drying protein glues. Also applying a new finish may call for some problems. A tendency for treated surfaces to darken is noted. And last but not least, don't use PEG on timber which has been steambent as this would risk a undesired extreme straightening.
To come to a conclusion, the restorers job is look at the pro's and con's of each possible treatment, and his first responsibility is to the object's integrity, not to the customer. Theoretically, the best conservation is always carried out by the owner of the object. He has something to put into practice as well. Not only is he in control of temperature and RH, but also can he change the placing of the furniture every once in a while. Just as a last example, a warped cardtabletop that is left folded open from time to time, can sometimes straighten completely by itself!
Thank you for your attention.