When weaving the lace shawl I noticed that the selvedge threads were considerably tighter than the rest of the warp. I couldn't work out why and assumed I must have done something silly when lacing on. With the shawl off the loom and having been allowed to relax and with a bit of pulling and stretching it seems to be OK.
I retied the warp with particular care to make sure that all of the ends were under equal tension. After about nine inches of weaving the same thing happened again - a tightening of the selvedges. The picture on the left shows how the warp is curving up at the selvedge. I looked at everything again, inspected my draft and then felt very silly when I realised the cause.
In balancing the disposition of the warp on the sectional warp beam warp I had beamed the warp so that I had the selvedges [four threads on each side] in their own section. These threads had more space so the warp did not wind up to the same diameter as all the other sections. This can be seen in the picture on the right.
As we all know, C/D = Pi, so the two selvedge sections that I measured as having a diameter of 5.9 cm has a circumference of 18.535 cm whereas the other sections with a diameter of 6.4 cm has a circumference of 20.106. What this means that for every 8 inches of warp released from the back beam only about 7.3 of the selvedge threads were being released. It soon mounts up!
It is all corrected now, the selvedge threads are off the warp beam and dangling at the back appropriately weighted. I now know that I should not have sections with less than the standard epi in them. This means a better balancing of the pattern repeat with the epi. If I need extra threads they must either be beamed separately [or dangled down the back] or I could fill the whole section anyway and just use the threads that I need.