<p style=”text-align: left;”>To add confusion, I tried to measure an oxy acet nozzle, it doesn’t fit any of my thread gauges, but my imperial ones only have even numbers at anything like the pitch. Interweb suggests 1/2 inch 27 tpi, another oddball.</p>

I think Harris is using a non-standard thread.

Nowhere in any table of American threads I have found, is 25tpi listed. The nearest is UNEF and even that is 3 turns out.

Further, one of its catalogue specifications promises a proper fit “if used with Harris fittings”. Oh aye?

Half-inch threads to US standards:

UNF 20tpi

UNEF 28

NPS / NPSF 14

NPT /NPTF 14.

I encountered this sort of problem years ago when I volunteered to make an adaptor to fit a brewery keg C02 cartridge. It was metric but its pitch matched nothing in the books. I forget what, and cannot now find my drawing, but many times-sums revealed a sufficiently close 2-stage change-wheel train for my EW lathe with 1/8″ pitch leadscrew and wheels in 5-tooth increments. The application? A warm-air breather, using soda-lime and CO2 in a heat-exchanger, for warding off hypothermia in cave-accident casualties.  I lodged a copy of the drawing with the rescue organisation but I believe they no longer use that device.

On 18 May 2026 at 19:16 Bill Phinn Said:

 

This thread may not be “standard” but it is obviously wanted enough for Taylor Tool to have it as a stock item.

I wonder if it’s down to the burners being designed specifically for propane, so folk don’t mix propane and acetylene components. i.e. it’s not a standard unified thread, but is standard for that gas use.
I remember the late J.S. discussing some weird Myford taps which were either imperial diameter with a metric thread, or vice versa.

Bill

I think I’ve got the pins at the short end the wrong way round. Other lever ratios might work better with a different pitch set, say 1.5 mm.

Messing around with https://bilar.co.uk/cgi-bin/change-gear-calculator.pl says a 32 and a 63 in the geartrain would give you 25.003

3D printed would likely work unless you envisage production quantities.

Edit, overlapped with Jason but fortunately we come to similar conclusions.

Edit again another solution only needs the 63T gear whcich as a common metric to imperical conversion may be readily available.

20 – 50- 80 -63 = 25.003 or 0.012% error

sorry it’s a bit faint. The end C of the lever is connected by a link to some fixed point. The point B is connected to the topslide via a clevis fixed in place of the topslide screw support. The end A is connected via a link to a post fixed to the cross slide.

Now suppose the lathe is set to turn 1.5 mm pitch. In one rev the saddle moves 1.5 mm (obviously), but point C doesn’t move, so the topslide moves by 1.5*BC/AC, and we want this to be 1/25″ which is 1.016 mm. Rearrange and we get BC/AC = !.016/1.5, so if AC is 150mm (to pick a convenient sounding figure) , BC is 101.6 mm.

As the saddle moves, the point A will move in and out  very slightly, which will make the link connecting A to the cross slide move from being not parallel to the bed. This will cause a very small error, which could be calculated if we knew how long the thread was, but it will be very small especially if you arrange the start finish points of the thread to have the ABC lever slanted each side  as in the second sketch. A similar error exists with point C but as the link that end can be much longer the error is even smaller.

Apart from an intellectual exercise I can’t see anyone actually doing this! On my lathe getting a suitable post in front of the topslide on the cross slide would not be easy.

Yes it would repeat the same every pass, and as I said it would be very small. All depends on lever/link lengths compared to length of thread. The longer the better for levers/links.