Are you saying that 94xxs were never fitted with screw reverse?

That seems to be the case, yes. Certainly neither 9400 or 9466 have it now.

RCTS doesn't mention the matter.

So presumably the story of loco sheds not liking the allegedly-fitted screw reverse and replacing it with a conventional lever is an old wives' tale.

Well, it would seem so. Its not impossible, I suppose, that some were screw fitted initially, maybe the first lot, but bearing in mind that screw reverse needs to interact with the firebox casing if not the tanks I can't imagine fitting it would be a shed job, even f the factory would supply the parts. I'll ask on GWSG. Someone there should know.

One could have a spot the difference here! Feel free folks, especially if you can spot something I'd otherwise miss when I come to my sketches. I was thinking that these two variants on the 79 class were so alike as not to be worth illustrating, but I've decided there's enough change there that it should be highlighted.

These are the Ahrons drawings, from the Holcroft Armstrong's book, of the 2nd Goods (1857) and the 4th Goods (1861). The thing that particularly struck me was the apparent lack of a sand box on the 79. I suppose there could be a tiny sandbox between the frames, (I've found what appears to be one such in "The British Steam Locomotive" but its a bit odd. The actual pipes are clearly different, Ahrons is not tracing his own drawings. It could be an error I suppose, even Homer nods. 

I'm beginning to find this early stuff rather fascinating, but always the way I suppose. I fear I lack the skills and perhaps more importantly the determination to follow @MikeOxon into modelling the period though. I will admit to being happy to have little in the way of locomotive breaks (we are in the 19thC here, breaks is correct) as I find them a particular pain to get right.

1 hour ago, JimC said:

I'm beginning to find this early stuff rather fascinating, but always the way I suppose. I fear I lack the skills and perhaps more importantly the determination to follow @MikeOxon into modelling the period though.

Thank you for the reference Jim.  The good thing about the early period is that there is very little information to prove if you've got it wrong 🙂 

On the few occasions where there are photos, however, I have found frequent discrepancies from the drawings.  It's an old problem:  as long ago as 1898, Sekon wrote in his book ‘The Evolution of the Steam Locomotive’ that “Readers may wonder why such obviously inaccurate statements should be published. One can only conjecture. Many lists of early locomotives have during the past few years been published. These should, however, be accepted with the very greatest caution.” 

Always try to find primary sources whenever you can and be prepared for later information to be inaccurate.

2 hours ago, JimC said:

One could have a spot the difference here! Feel free folks, especially if you can spot something I'd otherwise miss when I come to my sketches.

I suppose that the more significant differences are the presence of compensation beams on the earlier engine and the change from round-section coupling rids to flat rods in the later version. 

There are difference in the boiler and firebox supports, as well as in the smokebox wrapper. The spring gear looks to have been beefed-up in the later engine.  The reversing lever has changed and so has the spring balance but these could be something on the drawings, rather than on the engine itself!  I'm surprised that the leather buffers do not appear to have changed!  i assume that the vertical lines above the backplate on the later version are the edges of a small spectacle plate.

54 minutes ago, MikeOxon said:

On the few occasions where there are photos, however, I have found frequent discrepancies from the drawings. 

I suppose if no battle plan survives contact with the enemy then perhaps no arrangement plan survives the first general overhaul!

94xx reversing gear. I've been pointed at drawing 122690, which is for Lot 365 and is of reversing lever and quadrant. That's an October 1945 drawing for lot 365 , the Swindon build, so I'm pretty reasonably confident they were all lever reverse.

Its all very weird though. Freezer must surely have seen umpteen 94s. How did he get the idea they had screw reverse, and where did the other tales come from? Offhand I can't think  of any GWR classes that converted from screw to lever reverse in the right sort of timespan that the tales could be confused with. 2251s went the other way I believe.