Chris Knight
Established Member
I have recently become a bit paranoid about glues - blame the chair, I did not want it to fall apart!
I have conducted many tests now, generally crude in just about every which way but the pattern of results that emerges might be useful info for somebody. I should say however, that I am no expert and make no claims as to any real knowledge of the physics or chemistry of glues, so feel free to disregard everything I say!
My tests basically involved joining two planar faces of timber cut at various angles to the grain (eg 90 degrees- end grain, 0 degrees - long grain, mitred cuts 45 degrees for end/long grain) and joining these to other similarly cut or differently cut pieces, then after the glue had cured, breaking the joint apart to see whether the glue or the wood failed. Apart from the joining method (eg end grain to long grain), the tests also varied as to whether the joints had been specially prepared for gluing. For example, were they sanded, or planed in any particular way, were they wiped with a solvent before gluing and so forth.
In nearly all cases there was some wood failure. I guesstimated the failure I saw as a percentage of the joint area. Wood failure so observed ranged from 5% to 100% among the tests as a whole. I took wood failure to mean that I had achieved a good bond inasmuch it kind of represents a situation where I had solid wood rather than a joint - albeit solid wood of short or long grain with all that implies for strength.
So, wood failure was "good" and the oft-claimed benefit of XYZ glue "stronger than wood!" turns out to be true - WHERE THE GLUE STICKS!
Where the glue sticks turns out for me to be the crux of the issue. I found that any joint configuration including end grain to end grain - often said to be a bad configuration, can stick very well. As an aside, I should say immediately, I make no claims for longevity here, an end grain to long grain joint can be made to stick very well today but a couple of seasons of moisture exchange and the accompanying dimensional changes of the timber (which are impossible to stop completely) will render a good joint quite useless in time. This is why some writers say "EVERY GLUE JOINT WILL FAIL IN TIME" I do belive this but I also believe that the time involved can be very long if one respects the usual strictures on grain direction in one's joinery - hence my comments below on biscuits and dowels.
What I found was - the single most important factor in a good glue joint was to rough up the surfaces with coarse (100 grit) sandpaper just before gluing. This in no way creates any meaningful mechanical bond but it does two important things. First, it removes a good deal of grease and dirt that hinder gluing. Just handling your pieces after cutting them is enough to transfer glue-impairing oils or other contaminants from your fingers to the surfaces to be glued. It also helps remove natural oils as in Iroko that bleed to a fesh cut surface quite quickly. Second, it creates "hiding" places for glue that ensures not all glue is squeezed out if clamping pressures are excessive. The glue can still form its molecular bonds.
Wiping with acetone helps to glue difficult woods like Iroko and ebony but not nearly as much as sanding. Joint configuration is not very important to achieving a good glue joint if sanded first.
The most forgiving glue - in terms of bonding when conditions were less than perfect, was the set of PVA glues (eg Evostick, Titebond etc.) I think the fact that these glues contain water, which is an almost universal wetting agent, helps. Of the Polyurethanes, I found Balcotan to be far the best, Gorilla OK and Titebond to be useless. Of epoxies - to be used in the sort of quantity that say - making a table would require, Devcon 2 ton is the best. If needing more, then use the West System with suitable fillers (that ARE needed).
In short, I found that for 99% of gluing that Titebond in its various forms was best and they were all good but for indoor work where moisture is not a problem, I prefer the original. If slipperiness was needed for tight joints - as it was on my chair, then Balcotan Poly was best by a million miles.
All of the above opinionated findings are however based on "100 GRIT SANDED JOINTS" and this done just before gluing.
As to biscuits and dowels.
I see many folk using these for joining eg table tops. For reasons of strength, this is quite unecessary in that particular application - in my view and based on my tests. For alignment - well they can help and I won't argue, although I rarely use them for that. For joints specifically designed around their use as structural components ( and I use both biscuits and dowels for this at times) they are great.
I have conducted many tests now, generally crude in just about every which way but the pattern of results that emerges might be useful info for somebody. I should say however, that I am no expert and make no claims as to any real knowledge of the physics or chemistry of glues, so feel free to disregard everything I say!
My tests basically involved joining two planar faces of timber cut at various angles to the grain (eg 90 degrees- end grain, 0 degrees - long grain, mitred cuts 45 degrees for end/long grain) and joining these to other similarly cut or differently cut pieces, then after the glue had cured, breaking the joint apart to see whether the glue or the wood failed. Apart from the joining method (eg end grain to long grain), the tests also varied as to whether the joints had been specially prepared for gluing. For example, were they sanded, or planed in any particular way, were they wiped with a solvent before gluing and so forth.
In nearly all cases there was some wood failure. I guesstimated the failure I saw as a percentage of the joint area. Wood failure so observed ranged from 5% to 100% among the tests as a whole. I took wood failure to mean that I had achieved a good bond inasmuch it kind of represents a situation where I had solid wood rather than a joint - albeit solid wood of short or long grain with all that implies for strength.
So, wood failure was "good" and the oft-claimed benefit of XYZ glue "stronger than wood!" turns out to be true - WHERE THE GLUE STICKS!
Where the glue sticks turns out for me to be the crux of the issue. I found that any joint configuration including end grain to end grain - often said to be a bad configuration, can stick very well. As an aside, I should say immediately, I make no claims for longevity here, an end grain to long grain joint can be made to stick very well today but a couple of seasons of moisture exchange and the accompanying dimensional changes of the timber (which are impossible to stop completely) will render a good joint quite useless in time. This is why some writers say "EVERY GLUE JOINT WILL FAIL IN TIME" I do belive this but I also believe that the time involved can be very long if one respects the usual strictures on grain direction in one's joinery - hence my comments below on biscuits and dowels.
What I found was - the single most important factor in a good glue joint was to rough up the surfaces with coarse (100 grit) sandpaper just before gluing. This in no way creates any meaningful mechanical bond but it does two important things. First, it removes a good deal of grease and dirt that hinder gluing. Just handling your pieces after cutting them is enough to transfer glue-impairing oils or other contaminants from your fingers to the surfaces to be glued. It also helps remove natural oils as in Iroko that bleed to a fesh cut surface quite quickly. Second, it creates "hiding" places for glue that ensures not all glue is squeezed out if clamping pressures are excessive. The glue can still form its molecular bonds.
Wiping with acetone helps to glue difficult woods like Iroko and ebony but not nearly as much as sanding. Joint configuration is not very important to achieving a good glue joint if sanded first.
The most forgiving glue - in terms of bonding when conditions were less than perfect, was the set of PVA glues (eg Evostick, Titebond etc.) I think the fact that these glues contain water, which is an almost universal wetting agent, helps. Of the Polyurethanes, I found Balcotan to be far the best, Gorilla OK and Titebond to be useless. Of epoxies - to be used in the sort of quantity that say - making a table would require, Devcon 2 ton is the best. If needing more, then use the West System with suitable fillers (that ARE needed).
In short, I found that for 99% of gluing that Titebond in its various forms was best and they were all good but for indoor work where moisture is not a problem, I prefer the original. If slipperiness was needed for tight joints - as it was on my chair, then Balcotan Poly was best by a million miles.
All of the above opinionated findings are however based on "100 GRIT SANDED JOINTS" and this done just before gluing.
As to biscuits and dowels.
I see many folk using these for joining eg table tops. For reasons of strength, this is quite unecessary in that particular application - in my view and based on my tests. For alignment - well they can help and I won't argue, although I rarely use them for that. For joints specifically designed around their use as structural components ( and I use both biscuits and dowels for this at times) they are great.