Statistically speaking....
Brian in Sandusky, OH
>...it wasn't exactly a clinical trial.
They had three species (maple, oak, and ipe), and three joint fits (tight, snug, loose).
What makes this not-so-great science is the fact that they only did one sample for each type of joint and species combination (unless I read the article wrong). In order to really get statistical significance, they would have to perform the test on the same combination of wood and species many times and get an acceptable range and mean.
I wasn't so disappointed in the lack of clamping the joint. As long as the subjects were all treated identically, then it's valid in that regard.
And here's the big thing for me...on the bottom of the last page, they said that with the stronger of the glues, the failure was almost 100% wood failure, and not glue failure.
This being the case, then they are just proving the variability of wood from joint to joint, and proving nothing about what differs between the strong glues. This truly invalidates their data. If the wood is failing, then what information can possibly be extracted in regard to the glue strength and how each type of glue differs from the other? They were not measuring glue failure! What glue comparison can be made??? The only thing proven here is how a joint can differ in strength based on the qualities of the wood...NOT GLUE.
Now I'm not saying that the information isn't somewhat valuable, and I don't expect Fine Woodworking to become a clinical lab striving for statistically significant p-values, but I take the information with a grain of salt.
Oh well...not to overly criticize, it was still a good read.