Agree with DMS. I didn't/don't trust my eyes or knowledge enough to be confident with the look of ways (unless of course they are obviously bad). I am nubee, so following is not that of a seasoned expert, for sure.
I took a Starrett steel rule and a feeler gage when I looked at used lathes. Put the rule on the portions of the ways that wear (not the top of v-ways, but the angles and flats for flat ways) at the headstock and then checked with the feeler gage. Checked with 24" rule (a 12" might work also). Passed on the lathe where I could get .004 feeler in. Bought a lathe where I could only get .001 in. After cleaning, my 1954 circa Sheldon purchase appears in very good shape. Also, you can shine a good bright light behind the rule. The more light you see, the more space between the rule and the surface you are inspecting.
If you are serious about assessing a used lathe, knowing what I think I know now, I would have a good machinist (hopefully friend or supporter of a new would be lathe owner) make a true test bar of 1" or so diameter and 12" or longer with centers and take it with a dial indicator and magnetic base to your "prospect". Fit the test bar between centers (be sure to clean the bores and the centers both live (spindle) and dead (tailstock), set the dial indicator on the carriage and measure the test bar from head to tail. The movement (ideally, lack thereof) of the dial indicator will tell you if the ways are materially warn or if the tail is not aligned with the head (latter not a big problem, easily adjusted, far as I know). And by rotating the spindle/test bar, you can also measure spindle runout (a potential "used lathe issue" especially on old lathes with bronze bushings). Obviously, the rule and test bar must be true, and the lathe centers must be sound and fit correctly in both the test bar and the lathe.
I think runout/spindle wear is best tested by testing both horizontal and vertical slack by pulling firmly sideways (horizontal), resetting indicator and pulling firmly up/down(vertical). Alternatively, and absent a test bar, you can set up dial indicator directly on the spindle and pull/push on a Morse taper bit in the spindle or the live center if you don't have a MT bit. You can test the "no load" spindle runout by simply setting up the dial indicator and turning the spindle by hand, but when you are using the machine, you are applying side pressure and with bronze bearings (bushings) I think you might get an overly positive assessment using the "no load" test only. My Sheldon purchase had zero no load runout and +/- .001 horizontal/vertical when loading with a bit in the spindle. Maybe this relationship holds across more worn machines, I don't know.
If the lathe you are testing is not (reasonably) level, the test bar test of way wear may give misleading results. I don't think the spindle wear test(s) would be materially affected, but I am neither machinist nor used lathe evaluator expert.
I agree with DMS that way wear is almost undoubtedly greatest within the first 12 inches or less of the chuck. If the first 12 inches or so of the ways is not unduly worn, the rest will almost certainly be fine.
I haven't purchased a used lathe, but I have purchased a used mill. I think I did OK, and I used the advice from the page I linked. As far as what worn ways look like, aside from obvious physical damage (gouges, burrs, etc), worn ways are going to look more or less like good ways. the problem can be a deviation of only a couple thousandths, usually in the area within the first few inches of the chuck (where most work is done).
When I was looking at mills, I also found that cranking things back and forth to check for binding, and then grabbing them and giving a tug to check for slow was quite instructive. If the seller is trying to make the ways appear "tight" by cranking down the gibs, things are going to bind at the extents of travel if there is significant wear.
