I don’t think that constant tension is causing the problem, but rather a sudden increase in tension for a short period of time. That brief period of time that the loop is normally released from the hook to form the stitch. If so, it would appear that we may have a “thread flow” problem up top more than a thread tension problem. I think that the thread gets wrapped around the finger in the bottom when the upper thread loop makes a second lap around on the hook. Maybe a result more than a cause.
An increase or decrease of constant pre-tension should only effect how the upper tension assembly needs to be adjusted in order to form a proper stitch. Thread guide drag should remain constant until you add or subtract one. By adding thread guides as TnT has suggested and Tate has expanded on, might be doing more to effect thread travel and thread control rather than thread tension. It still may have a lot to do with thread twisting, which thread moisture may play a significant part.
The problem is, when the event occurs, the operator’s eyes are usually on the needle area and not watching what’s going on up top. If twisted thread not moving through a thread guide or the tension assembly causes the loop to not release from the hook, the second lap around the hook will jerk the twist past whatever it was hung up on and no longer be there after the event.
A similar event is when the thread gets caught in the stow slit on the old spools, except in that case, the hook can’t jerk the obstruction loose, so the thread breaks instead. I still think that the flow of the thread up top is our culprit, and that the humidity in the air and moisture in the thread has something to do with it. Even the change in thread manufacturing processes over the years could be complicating the diagnosis by some thread acting one way on these old machines, and other thread acting differently, depending on when and by whom it was made.
CD in Oklahoma