I've noticed something unexpected while testing PVA glues for strength.

Glues tested: Titebond Original, Titebond III Ultimate.

Wood species: Beech and Pine (and other unknown hardwood species).

Transverse to transverse grain joinery. All mating surfaces sandpapered.

Control: Glue applied evenly to both surfaces with brush then clamped straight away for 24h.

Test 1: Glue applied evenly to one surface with brush, then left to dry 1 hour. Then glue applied evenly to both surfaces before being clamped 24h.

Test 2: Glue applied to one surface with brush, excess glue scraped off with sandpaper (leaving surface moist), left to dry 1 hour. Glue applied evenly to both surfaces then clamped 24h.

I had expected the control joint to be stronger than test 2, and test 2 to be stronger than test 1.

In reality, the joints would seem to be equally resistant to sheer stress and longitudinal traction. All joints had barely visible glue lines, the difference between test 1 and the others is in the range of microns suggesting most of the dried glue had been absorbed into the wood during clamping.

The results seem to be reproducible, meaning this could have implications for correcting errors in assembly, or repairs and maintenance of cured glue joints.

Do you think this is unusual, given what we know about how PVA glues work?

  • 3
    "In reality, the joints would seem to be equally resistant to sheer stress and longitudinal traction." You have scientific equipment that measures that ? How do you plan to effect the entire woodworking industry with your findings ?
    – Alaska Man
    May 10 '20 at 18:06
  • 2
    Yes the conventional wisdom is wrong, for a given definition of wrong (details matter). As for your tests, what are the actual strength results you got? The key thing to remember in anything like this is exactly what you're aiming to match — match only, you cannot exceed it — so unless you are matching the strength of glues used normally the results are really of limited to no value in practical circumstances. Regardless of this, if your results are of equal strength then you're not clamping hard enough, simple as that (i.e. you're not achieving the maximum possible strength from the PVA).
    – Graphus
    May 10 '20 at 18:30
  • 2
    Your testing methodology has exposed a flaw, as you suggest in your own answer. PVA glues, once cured will tend to not form a good bond with fresh PVA glue. A full cure is dependent on a number of factors you might be able to control for, but 1 hour is probably not long enough under most conditions.
    – jdv
    May 11 '20 at 0:14
  • 2
    There very well may be some conditions where the received wisdom does not hold true, but these wisdoms are often practical simplifications in a complicated world. I'd look to what folks who do a lot of regluing have to say about practical experience. Furniture restorers and luthiers have a lot of experience with a variety of glues, and how to glue up previous glued surfaces. I know most guitar makers would remove cured PVA from surfaces if they wanted a guitar not to fold up in half under stress.
    – jdv
    May 11 '20 at 0:17
  • 4
    I suspect that 1) conventional wisdom is correct, and also 2) what you found is true insofar as the joint is "strong enough", and 3) often we don't need the strongest possible glue joint. This is true a lot in woodworking; we don't need the "best" joint, we only need one that is good enough. Or said another way, what is "best" is dependent on context. All other things being equal, you want the strongest glue joint possible. You definitely want one that won't fail in use! Whether your method achieves that, I don't know. It might! I don't think that invalidates conventional wisdom, though. May 11 '20 at 5:44

In the absence of helpful comments, I've done some research on PVA glue chemistry to see what this could be. The general rule of thumb is that wood glue does not stick to cured wood glue.

Wood glue binding undergoes two phases.

  1. A short-term drying phase (< 2 hr), during which the material is simply dehydrated.
  2. A curing phase (24 h+), in which electrostatic and covalent cross-links are formed.

Source: Ülker O (2016). Wood Adhesives and Bonding Theory. DOI: 10.5772/65759

I'm no expert on PVA glues, but what I'm observing could be as simple as partially dried surface glue being mobilized again through re-wetting before cross-linking occurs.

This would then allow it to be absorbed into wood or cross-link with glue on the adjacent surface.

  • 1
    Also, make sure you provide a reference for that image.
    – jdv
    May 11 '20 at 0:08

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.