Knot size
The main advantage of learning to braid all the possible knots instead of a few 4 and 6 bight "one size fits all" old standbys is the ability to create a knot that truly fits it's place "on the work".
Diameter or circumference
In the templates and photos for most of the examples in the topics covered so far I have shown the knots with an open lattice structure as the one on the left. In fact in the section on mandrels I have reccommended braiding the knot with this open structure for ease in working and leaving space for a two pass interweave if desired. ( Yes I am using one of the "old standys" here for simplicity but the principles apply to any odd part knot with any number of bights.) Another thing to notice is that the "pitch' or angle of intersection of crossing parts is 90 degrees.
When the excess lace has been removed by tightening the knot to close the lattice structure the knot will appear as seen in the right hand template.(Actually it is just almost closed. If it was really tight there would be no white space at all.) There is a specific circumference around which the knot will tighten and keep this exact form which may be approximated by the expression 1.4 X the width of the lace X the number of bights in the knot. For 1/8th inch in this 4 bight knot that would be 1.4 X .125 X 4 = .7 inches. This is probably close to the minimum circumference for a good appearance for this particular knot with this lace width.
The photo to the left is the same 5X4 knot done on a 1/2 inch mandrel with 1/8th inch roo lace. This leaves an open lattice for working room.
The photo to the right is the knot snugged onto a 1/4 inch dowel. This is close to the calculated minimum circumference from above. It is approximately 1/2 inch long.
Here on the left is the same knot tightened on a 1/2 inch diameter dowel. This is twice the circumference of the one to the right above. It is still appx. 1/2 inch long.
The photo at the right is on a 3/4th inch dowel. Now it is about 1/16th inch shorter here. Notice that the span of the bights is getting a little bit longer than we would really like for appearance sake.
In summary
I used the simplest basic knot in the illustrations but the same principles apply to any size knot.
First, the technique of using a mandrel of a size that will gives an open structure for the knot allows us to put the knot on just about any "work". That is on round braid, strap work, and other irregularly shaped items.
Second; A knot with a particular number of bights in a given lace width will "fit" a range of three to one in circumference with a minimal change in length. Probably the best appearance for the knot will all in the middle of this range but while the number of bights ultimately determines the relative size of a knot, it is flexible to this extent. In other words, due to the flexibility of the structure, a knot with a given number of bights is not one exact size, but will fit a limited range of core sizes quite well.
Length
While we have a certian flexibility in the circumference, the length (or width if you wish) is much more dependent on the number of parts and the width of the lace. This is the basis for my personal preference for using knots with an odd number of bights. By using prime numbers such as 5 or 7 for the bights the increment in the parts for the possible knots is is greater in number and in smaller steps. (Look at a chart of the possible simple turksheads to see what I mean by this.) In the range from 2 to 13 parts, there are only 5 possible knots with 4 bights while there are 8 for 5 bight knots and 10 for 7 bight knots. Also all the small quick start knots have an odd number of bights.
Since each part of the knot is crossed by another part, the number of code rows in a vertical line are equal to one half the number of parts. Thus the maximum length of the knot is 1.4 X 1/2 P X lace width(use the next whole number for odd part knots). When the knot is tightened to a larger diameter than minimum the length will shorten slightly.
Because of the flexibility of the lattice structure of the braid, these computations are only approximate. I don't know of any way to get exact measurements without actually braiding the knot, tightening it in place on the work and measuring it. However these figures will give a "ball park" estimate for a starting point for a trial and error attempt and with a little experience will prove adequate for the purpose of selecting a knot to fit a given job.