Keri & Bill, I am very lucky to have access to a 3D scanner. We even have a 3D plotter at work that "plots" solid objects out of resin. Great for prototyping and evaluating ideas before progressing to the final manufactured item. Fortunately I do not need this technology on my gearbox project, a 150mm vernier (accurate to 0.02mm), a 30cm steel rule and a set of feeler gauges are the only tools that I have used to measure up my box. Oh, and of course, a micrometer to measure shaft diameters. For my purposes I really don't need to know (accurately) the dimensions of the original gears (so long as my new ones clear the selector shafts, casing, and other gears around them). Its primarily axial dimensions along the length of each shaft that I am concerned with and (fortunately) the thrust clearances (and their tolerances) are well defined in the S&R manual.
As for the gears, the dog rings, the selector rings etc - well that's all a new design (no reverse engineering involved whatsoever). All I am doing is re-utilising the space around the existing shafts to the best of my ability (making gears as wide and as strong as possible etc) and this requires that I undertake the required engineering calculations in order to make sure that my design is sound. There is no need (fortunately) to simply copy the original dimensions and hope like hell that I measured them correctly for fear of something binding or seizing.
The moral of the story is: If this gearbox fails its because I'm a bad engineer, NOT because I don't know how to use a vernier and micrometer.
Mark, the box should be good for 500 to 600hp, but this is by virtue of the fact that straight cut (spur) gears are much more efficient than helical gears (as used in standard road going box). What this means, ultimately, is that the amount of heat that is put into the box is reduced - and hence the thermal loading of the oil is reduced. So this figure is not an indication that the gears are necessarily stronger. Indeed, the figure that defines the strength of a gear train (a chain only being as strong as its weakest link and all that) is the static failure torque (applied to the input shaft) with the output shaft locked. This value is scarily low on many gearboxes, but (fortunately) represents little else other than the amount of torque required to break a gear or shaft (whichever goes first) and has little relation to what happens on the actual car since the object of your gearbox is to
transfer your engine torque to the wheels with as little torque dropped over (and hence work done upon) the gears as possible.
What will ultimately define the lifetime of my gear box though is the amount of backlash between the teeth on the dog-ring and the teeth on the selector ring. Look closely at the above pics. I have a mark-to-space ratio on my dog teeth of about 1:3. Can you picture what will happen in my box as I take my foot of the throttle and those teeth bang back and forth? Those teeth are going to get the crap kicked out them and will likely require regular maintenance. That, unfortunately, is the price you pay for this type of gearbox. FYI, On a road-going box the mark to space ratio is 1:1, which is why a synchroniser mechanism is required, else you'd battle like hell to change gears (ever crash shifted your way home after losing a clutch?). You effectively don't need synchro-rings on a crash box because of the large mark-to-space ratio. A 1:3 mark-to-space ratio gives exactly a 50% probability of a flawless shift, no clutch required.
The chamfer on the edge of the dog-teeth takes care of those occasions when the shift isn't quite so flawless
The number will be duly applied to the sides and rear of the car.
So all in all, perspex is NOT a good idea. 
Oh well, we learn by trying these things 
