Monday 9 July 2018
The received wisdom on gluing end grain has often been to avoid it altogether – not only end to end grain but also end to long grain gluing. Of course there are reasons for this, but many practical joints depend at least partly on the strength of glue bonded to end grain, so I believe there is merit in exploring the issue.
In this article I will look at examples of where end grain is glued and then use some simple tests to explore how far confidence is justified.
When glue is applied to wood it must wet the surfaces to make continuous contact. As the glue sets it makes a chemical bond with the wood fibres. The joint forms a sandwich with a thin layer of solid glue in the middle and a thin glue/wood mixture either side. Outside of this is the unadulterated wood.
Ernest Joyce discusses end grain gluing in his book The Technique of Furniture Making and explains that 'The cut pores of end grain will suck up glue like a sponge, preventing any permanent bond'. He then describes other problems and the traditional method of sizing end grain with hide glue.
How strong is it?
We sometimes hear that a glue is 'stronger than wood itself' and while this is generally considered to be true for most modern glues, it does beg the questions 'which wood?' and 'in what direction?'
Anyone working with wood quickly finds it immensely strong along the grain direction but relatively easy to pull apart sideways or split with the grain. Bruce Hoadley's book Understanding Wood tells us that pine (Pinus spp.) is 40 times as strong in tension along the grain as across it, while weight for weight some hardwoods along the grain are stronger than structural steel!
It is not surprising then that simple butted glue joints cannot match wood's long grain strength while they easily match its strength across the grain. But does this mean that a glue bond to end grain is weaker than a glue bond to long grain? Some simple tests described later in the article may cast light on this question.
The joint between the neck and body of a violin has tension from the strings continuously trying to fold it forwards on itself. In addition to this there are moving forces in all directions from the player and vibrations strong enough to fill a hall with sound! Even the slightest movement between neck and body would render the instrument useless and yet this joint consists of a shallow tenon with very little long to long grain contact. Most of the glued area is between the end grain of the neck and the long grain of the internal top body block.
It would be appealing to think that well made joints relied entirely on mechanical strength and the adhesive is only there to stop them sliding apart, but that does not take account of dynamic loads. Any small mechanical movement in a joint under intermittent pressure causes fibres to be crushed allowing larger movements when the pressure is repeated. This means that unglued joints tend to become sloppy with use.
Consider for example a mortise and tenon chair joint. Heavy use would cause an unglued joint, even a very tight one, to work loose thus allowing the chair to rack. Glue forms an important part of a working chair's strength in terms of keeping the joints rigid so flexing can only occur in the rails, which will deform elastically and spring back.
In a chair joint, each tenon has four cheek surfaces and four shoulder surfaces that may be glued. The shoulders on the tenon are all end grain and two of the cheeks contact end grain in the mortise. That just leaves two cheeks with long grain to long grain contact, and these have the grain running at right angles so any wood movement could potentially break their bond. The tenon, in this instance, therefore benefits from end grain gluing.
Conventional drawers are made with extra thin sides giving the lapped or half-blind dovetails joining them to the drawer front a small area of long grain to long grain contact. Most of the glued area on the faces of the tails is bonded to end grain in the socket. As the drawers are slid in and out â€“ which is not always as carefully as they should be â€“ they are subject to racking forces resisted by all the glued surfaces in the joints. As most of these are made to end grain the dovetails also benefit from end grain gluing.
Sizing end grain
Traditional hide glue, which is always used by violin makers but rarely used by contemporary furniture makers, has several unique properties. One of these is the ability to form a strong bond with previously set glue, which means it can be applied in layers. This means a thinned down layer of 'sizing' can be applied to end grain before gluing to it.
However, modern adhesives will not bond well to previously hardened glue and so traditional sizing is not practical. Instead I suggest flooding the end grain with an excess of glue and rubbing it in to saturate the cut ends of the pores as far as possible while gluing up the joint.
Pine and PVA
A few years ago I did a simple test on the strength of butt joints between end grain and long grain pine using a modern PVA type adhesive.
After the glue was set I broke the joint and found that it failed in two ways: one side suffered glue failure with a thin layer of glue left on the end grain. The other side suffered wood failure, tearing the fibres out of the long grain leaving them firmly attached to the end grain. This looked interesting because it suggested the wood and the end grain bond were similar in strength.
Oak epoxy joints
Recently I repeated the end grain gluing test, this time making various configurations of test samples using oak and epoxy resin adhesive. Unset epoxy resin is viscous and water-free so it cannot be drawn into the cut ends of wood fibres and I thought this might be a better adhesive for testing on end grain.
The setup with a 20 litre bucket hung on a rope to apply increasing loads to each joint was crude but fairly repeatable. Pouring water slowly into the bucket I increased its weight at the rate of 1kg per litre of water. Lever action would multiply the force at the top of the joint five-fold.
I decided to compare sample joints made near the end of the long grain with those made near the middle, thinking that the fibres running along the edge of a board would be easier to tear away from the end than from near the middle.
In these cases it wasn't the glue bond to the end grain that failed but the long grain of the oak itself. The force needed to break the joint was similar each time but the joints that I made near the end of the long grain tore much deeper into the wood than the ones near the middle.
Joints made between end grain and end grain broke under a similar load but this was always due to glue failure rather than wood failure.
Clearly I am not proposing to make corner joints in furniture using just glue, but I think it helps our practical understanding to look more closely at what happens when glued joints fail.
Firstly these tests emphasise the well-known fact that wood fibres are immensely strong longways but relatively weak from side to side. More surprisingly they indicate that side-by-side fibres are the source of weakness in simple butt joints, rather than the strength of modern glue bonds, which I found to be similar whatever the wood direction. It follows that within a traditional joint, glue applied to end grain can make just as good a contribution to strength as glue applied to long grain. It would be interesting to hear other people's findings on this subject.