Nicely illustrated … and yes, it looks serious

MichaelG.

This is a bit lengthy to properly explain.

I don’t automatically trust an accuracy specification for any tool or gauge. You have to start from a known base line and work from there. Better would be to first start with checking your test bar by measuring along it’s length with a very good micrometer while checking for any variations in it’s diameter.  Then indicating it in a pair of V blocks while checking for straightness would be the first checks I’d do. If it does measure as being parallel, then rotating it in a pair of V blocks and indicating at various points to locate any areas of inaccuracy in its straightness and how much out that may be. Yes in an ideal world we wouldn’t have to check, I’ve also found in a brand new set of “certified” as accurate to low millionths of an inch gauge blocks there were two identical blocks that were marked and measured the exact same size. Mistakes or even defects can turn up with even the best manufacturing and quality control no matter what the supposed guarantee of accuracy might be.

Morse taper test bars no matter how well made and accurate they may or may not be also depend on an almost perfect condition female taper. Any previous internal damage, rusting, or even the slightest speck of missed dust or contamination makes the test results while using one just about worthless. Frankly and in my opinion, most hobbyist’s don’t or aren’t willing to look after those high precision head and tail stock tapers as well as they should. And at one time I was certainly no different. I’m also not against using those test bars, I have them myself. But I’m very careful to check any Morse Taper there being used in first.

In one way or another, the worlds machine tool manufacturing industry still use the same methods and procedures developed by a Dr. Georg Schlesinger that can be found in this PDF. https://pearl-hifi.com/06_Lit_Archive/14_Books_Tech_Papers/Schlesinger_Georg/Testing_Machine_Tools.pdf

And despite what some seem to assume is correct, the better built lathes even when brand new are not aligned to face parts perfectly flat for a couple of very good reasons. Properly set up, its desirable to always have them face very slightly concave and never convex. That way two mating part surfaces joined face to face will sit flat and not rock. And a lathe first set up to face dead flat from new would begin to wear towards facing convex as the cross slide ways start to inevitably wear. Secondly and on the better lathes, there’s usually some misalignment purposely added to attempt to compensate for possible large work piece weights and cutting forces. In general, those would show the head stock and also the lathes tail stock to be pointing up and also towards the operator by approximately .001″ over about a 12″ length of extended test bar. Not all lathes are set up this way, but in general the more expensive ones would be. The cheaper and more hobby level lathes today might be anything but correctly aligned unless they do come with a certificate of accuracy. Even then I still verify those numbers myself.

Afaik there’s two different methods used that vary between manufacturers to force a lathe to face concave. Some miss-align the head stock, while some bias the cross slide way surface grinding and / or scraping to do the same. Without exact details to fully prove it, I believe that cross slide bias may be more of a North American method, and very slightly twisting the head stock towards the operator is more of an European method. Although I suspect that isn’t universally true in all cases. Only my personal opinion, but for a few reasons I think the European method is still the better choice.

However it’s done, that introduced concave facing alignment isn’t very much. From memory and depending on the grade specification of the lathe, around .001″ for something like a tool room lathe -.002″ maximum for something like a more general purpose lathe over about a 24″ part diameter. Which doesn’t help us with our much smaller swing lathes. Since a lathe cuts on both the front and rear of the part as it’s facing, that 24″ diameter face is in reality a 12″ facing cut. So the misalignment to face concave is only around .000083″ – .00017″ per inch of cross slide travel. At those numbers it’s getting fairly hard for what most of us might have available to even measure it.

Impossible it’s something new as without question others would have come up with the same method and long before I was even born. Although I’ve not yet seen it mentioned anywhere else. But I did figure out a way that our smaller swing lathes can be checked, and a bit easier to verify by doubling those very small numbers.

You’ll need either a lathe face plate or a short and fairly thick metal disc that’s just about the largest diameter your lathe can swing. How and what you manage to use for the test cuts is up to what you have available and how much effort your willing to put into this. Secondly at least a .0001″ or it’s metric equivalent reading dial or dial test indicator. Your cross slide gib should be properly adjusted to allow a smooth but shake free travel. On a highly worn lathe that’s also going to have a direct effect on your measurements. Your lathes head stock bearings should also be in good condition and correctly adjusted for there specified radial clearances and end play amount.

Make a clean up facing cut across that face plate or disc. Now make a fine and steady facing cut maybe a few thou deep across the whole face with the very best surface finish you can manage. Then withdraw the cross slide back towards you. I’ll use a clock face and it’s numbers for explanation purposes. Looking from the tail stock end of the lathe at the object you just faced, 9 o’clock would be the left edge of the face plate or disc, and 3 o’clock would be the right side or furthest away from the operator during normal lathe operation. Set up and zero the indicators dial at 90 degrees to the part face about to be measured, and in the case of something like a face plate, just past it’s center hole. Slowly and gently move the cross slide inwards with the indicator tip traveling towards the 3 o’clock position on your face plate or part disc.

What that does will then show double the measurement of what the cutting tool removed to produce either a concave, dead flat, or convex facing cut. If you set up your indicator to travel the same path the cutting tool did, it will tell you nothing of importance since the indicator tip is then duplicating the exact same path the cutting tip did. With those measurements, then make your head stock adjustments accordingly until it will produce that concave facing cut within the above numbers. It takes almost as long to explain the how to as it does to do it.

This is also a very good example of never touching a lathes head stock unless you do fully understand what your doing and what the actual alignment numbers ought to be first. However the test method I’ve detailed can still be used to verify any new or used lathe. But it still won’t tell you everything, worn cross slide ways could easily indicate the head stock is out when in fact that wear should be the item to be addressed first. So some proper judgement about the lathes condition and amount of wear needs to be used as well.

That Dumbell or I’d assume the Two Collar test method was what was meant is simpler, but part deflection can also affect the test numbers. And with highly worn or a less than perfectly straight lathe bed, it also won’t tell you everything about the head stocks alignment. Doing a very large facing cut on a face plate or disc will at least provide direct measurements of an issue or misalignment with either the head stocks position on the ways, or an indicator of an undesirable amount wear on the cross slide way surfaces. It then depends on the lathe condition itself for what needs addressing. Lathes are fairly simple machine tools, but there alignments aren’t because theirs so many individual parts directly related to there measurable accuracy for any parts being produced on them.

This is why I wouldn’t buy a lathe that has new paint on it. The lathe was probably built on a jig in the factory.

The endless debate I referred to was on forums such as this, not among manufacturers, who obviously know their stuff.

They don’t set lathes to face concave. They set them to face non-convex. IE 0 to .001″ concave. The zero end of that tolerance is dead flat, not concave.

And if, hyporhetically, you set your headstock alignment to face .001″ concave over say a 4″ radius  , that would give you a .001″ taper over 4 inches when cylindrical turning. Which is why headstoch alignment is not used to adjust facing tolerances. Cross slide alignment is.

And the debate goes on, and on…

One can hypothesize on what to do with this lathe until the cows come home.

All the measurements above are being taken off faces that have been subjected to wear.

The only portion of the bed which will not have worn is under the Headstock or the extreme end of the Bed at the tailstock end. This would be the Zero dimension. This is of course assuming no-one has tried to rectify this lathe before?

A straight edge the length of the bedway and feeler blades will ascertain which of the vertical faces is the straightest and least worn. All measurements are then taken from this face to get a true picture of the wear.

This bed could be rectified by scraping, but that is a tedious job at best. The easiest course or action would be to grind the top face and kiss-grind the vertical faces at the same time. This only being done after first grinding the Bed mounting faces with the bedways on the grinders magnetic table.

Regards

Gray,

On 9 February 2026 at 15:30 Sonic Escape Said:

Now I’m waiting for a new belt, some gearbox oil and some steel and aluminum bars for testing. I want to see how accurate the lathe really is.

I managed to reassemble everything except the lead screw. Looking at the tool post I would say the previous owner was not lacking in creativity.

 

On 8 February 2026 at 17:10 Dave Halford Said:

This is why I wouldn’t buy a lathe that has new paint on it. The lathe was probably built on a jig in the factory.

I painted it 🙂

I knew that, you shouldn’t have with out the assembly jig. Enjoy (O:

On 15 March 2026 at 09:35 southernchap Said:

The truth ultimately is in the cut obviously, but I’m surprised there’s been no mention of Rollie’s Dad’s Method (RDM) for checking alignment.

I would rather check the lathe actually cuts parallel.

Tony

I found that the Atlas 12 x 24 that we have at the museum has an acceptable headstock alignment, but the MT3 on the spindle was not very good when checked with a MT3 test bar, even though there was nothing particularly visually wrong with the internal taper. I ended up by taking a tiny skim to clean up from the bore with a solid carbide boring bar with an aluminium insert. The taper was too long for the stroke of the Atlas top slide, so I set up the spare NOS Smart & Brown model A one on the Atlas temporarily and successfully improved the alignment of the taper. Arbors sit about 3/16″ deeper now.