The Reef Knot (sometimes incorrectly referred to as the Square Knot) is unreliable and unsafe as a bend. It should never be used for that purpose.
It's perhaps the best bend for rubber bands, and has been used and recommended
for webbing/tape by climbers/cavers (French?)--I was shocked to see such in some
in-use tape, but on playing around with it find secure (& asymmetric).
The Sheet Bend and the Double Sheet Bend can be useful for tying ends of cloth together, but they are generally considered to be too weak and unreliable for rope or string.
?! These knots are in common use, often with some end-securing mechanism,
and often in similar becket-hitch forms.
The Carrick Bend is generally considered to be strong and secure,...
... though w/o any clear data to support this--just the parroting of knots books!
... but it might be difficult to remember how to tie it properly if you don't use it very often. It is displayed incorrectly in a number of books and websites, and it can be very unsafe if it is not tied properly. It is usually very easy to untie, which makes it a good bend for string.
There are various things shown under the label, and perhaps more often so than for
other knots, yes. As for string, I find it rather fiddly to tie the lattice-form, and am
not much practised at any other.
The Ring Knot (Water Knot, Tape Knot) is easy to remember because it is simply a rethreaded Overhand Knot, and it results in a nice-looking knot. It is fairly easy to tie and dress, and it is a fairly strong and secure bend, but it can be difficult to untie after holding a load. It is also recommended when tying together webbing (or tape), wet materials, rubber materials, and so on.
There is a difference in knot-form for rope (round material), intended to be symmetric;
in tape, the commonly shown knot is asymmetric, with
interior/exterior sides,
though the knot can be arranged to be symmetric (which makes it a tad longer &
less *fat*). (Alas, indications are that the part that breaks is the exterior SPart,
and the part the slips in some stiffer tapes is the exterior end, which implies
choosing between strength vs. security (in theory).)
The Figure-Eight Bend (Flemish Bend) is fairly easy to remember because it is simply a rethreaded Figure-Eight Knot, but it is not as quick to tie and dress as the Ring Knot. It is also bulkier than the Ring Knot. Due to its twists and turns it has more internal friction than the Ring Knot, and therefore it is stronger and more secure. For this reason, climbers often prefer the Figure-Eight Bend (with some type of Stopper Knots) over other bends for situations when lives are at stake. The "Abnormal Figure-Eight Bend" is tied like the Overhand Bend, except that a Figure-Eight Knot is used instead of an Overhand Knot. The general consensus among climbers seems to be that the "abnormal" version is not a very reliable bend. For example, when the ropes are pulled in opposite directions then it can cause this knot to distort and become unsafe
It doesn't work so neatly in tape. I'd say it's stronger for being able to retain less
sharp bending of the SPart, but YMMV on data on relative strengths: AMGA (?) opines that
the Fig.8 & Ring Bends are equally strong. But, they don't show WHICH form of either
bend is actually tested, and of course not how it was set (assuming
perfect forms were
used for both--often the Fig.8 is tied asymmetricly).
Prefer "Offset" to "abnormal"--the former objectively descriptive, the latter not
necessarily so (the normal way to join rappel ropes is with an abnormal Ring Bend,
by such nomenclature--"Offset Overhand Bend" is preferred (or "Offset Ring...")).
And one canyoneer continues to use the Offset Fig.8 well aware of some of the climbing
troubles with it--indeed, having seen the ropes of one fatality (Mt.Zion), which he
described as "cables" (i.e., quite stiff).
But the warning I think should be given; the alternatives are readily available!
... However, the Alpine Butterfly Bend has some potential advantages over the other three bends ...
Hmmm, I don't want to chase all of that rationale, but tying a knot correctly
isn't all so
hard for the others. Some of the tying methods I've seen for these knots are the
problem, getting two clever by half, IMHO. I start with one Overhand and reeve the
opposite rope into it accordingly. Now, the PABPRES has a whiz-bang way of tying
the Butterfly which gets him a hotter first sip of his coffee, fine. In the course of
regular usage, one can get good at whatever method. (-:
Most bends should only be used with two ends of rope that are about the same thickness (which bends are exceptions to this rule?).
The OOB (Offset Overhand Bend) used for joing rappel ropes will often be called upon
(or could be) to join diff.sized ropes (7mm + 10mm, say); there is a particular
orientation that should be used here (thinner rope
outermost).
Sheet bends & derivatives of them & Lapp Knots are useful in thick-thin unions.
Other bends can be adapted sometimes with minor variation.
Flexibility & surface friction are important characteristics of "similar" along with
diameter--often more so, if dia. diff. isn't great.
--dl*
====