At Clausing's suggestion, I tightened the spindle bearing take-up nut, which adjusts the preload on the spindle bearings, by about 1/4 turn.
The headstock now warms up if the lathe is run constantly; previously, there was no perceptible temperature change. In retrospect, the failure to heat up after running for say 20 minutes should have been sufficient clue that the preload wasn't sufficient.
Tightening made a huge difference. I can now just about manage Clausing's original high-speed smooth turning test, although the lathe is still right on the verge of chatter with the chuck. It turns out that higher speeds chatter less, probably because one can get above a major resonance. The chips came off hot and blue. The depth of cut is
0.125" (0.25" diameter reduction) and the feed is 0.0026" per turn. I think I had to back off somewhat on the depth of cut, but it was in the range.Then I went back to the 5C collets. Now I'm able to take a 0.250" depth of cut (0.5" diameter reduction) out of a 1" diameter 1018 bar at 1600 rpm, in power feed (0.0026" per turn). The chip spirals off steadily, turning dark blue as it goes, the oil emulsion coolant boils on the workpiece, and the cut surface is a mirror.
The tool is a BXA-16N with pos-neg carbide inserts.
I was also able to cut the 1" bar off at 600 rpm in direct drive, with the SGIH 19-2 blade upside down and the lathe in reverse, a heavy spray of oil emulsion coolant, and a power cross-feed of 0.0013" per turn. I generated a spiral chip at least 6' long. This chip remained steel gray, and was not hot.
I'll have to try the BXA-7 cutoff tool again. Maybe it works now.
It seems that just about everything that is capable of being loose was loose on this lathe, and one by one I've found and tightened and/or replaced the loose parts. I don't think that this lathe saw much cleaning and adjusting, or preventive maintenance, and things quietly degraded over the years. Probably, people just shrugged and said it's old so what do you expect, and never thought to ask anybody who knew anything about machine tools if it was necessary to tolerate these problems. Or they were too cheap.
In any event, this lathe is beginning to act like a real lathe, and I bet a lot of things that didn't quite work right before will now suddenly work a lot better. This looseness had to have been undermining everything to one degree or another.
Joe Gwinn