Question for Jeff regarding forming a flat piece of metal over the compound curvature. Would you start at the bottom where it was flatter and work up or start at the more curved top and work down?
Good question. Of course none of us can know what the installer 'would' have done in 1937, but -
After drilling a few locator holes around the periphery (top, bottom and forward edges - from middle toward, but not to, the corners) and installing a few
clecos to position the sheet, I'd probably tack the middle of the lower and forward rows with two or three rivets first, then careully work the top row from the middle with 3 or 4 rivets, trying to nudge the metal into the desired contour as much as possible.
As to the contour - working this flat sheet onto that skin as I envision, the aft edge (notice was last) is going to need to absorb a bit of 'surplus', e.g. 'bubble' effect: you are not going to get a great deal of 'contour' out of this as one would a
hydroformed or
wheel worked metal, but some effect toward that so as to make a graceful match to the bird's natural shape. Hence one reason I've viewed the stiffeners as desirable. I could be wrong: the installer may have simply contoured the covering in one dimension, a simple roll shape. Given the area, that could also have been done without too much loss of aesthetic, but it's an even shorter cut in my view than what we see in 2-2-V-1 as a field effort, given the shape of that bird. Even so, stiffeners would be needed all the more on a flat section to resist oil canning, I believe.
To help stabilize the metal into that compounded shape, I'd start working the stiffeners onto the covering and tack them much the same way, start at middle with 3 or 4 rivets / skipping to about every other hole in doing so, and would probably tack the forward ends with a fastener or two. What we see in 2-2-V-1 are very neat rows - straight, and with rivets evenly spaced. We also see some slight but odd convergence / divergence of these rows: as neat as they are individually, for some odd reason some converge slightly going 'aft' (as we envision the installation), and others diverge slightly. This may be important as it suggests, to me anyway, that while care was taken to make the installation attractive, more focus was given making the metal behave in terms of where to place these stiffeners. In any case, they do not appear - by their comparitive spacing to each other, to match a factory pattern: this is in part why 2-2-V-1 is so intriguing as to what it may just be.
Once the stiffeners were underway, I'd look at how the cover was shaping up and and start tacking the aft row - just 3 or 4 rivets, to start. If trying to work a convex compound curve effect into this sheet, I'd evenly space about 4 fasteners and leave a slight bit of bulge between them to even suck up the surplus stock. This 'bubble' would have been set by tacking the upper row, after having done the bottom and forward rows.
BTW, to get an idea of what I'm trying to describe in terms of 'bubble' or contouring, lay a piece of paper on the table and tape three edges, taping the third edge so as to leave the fourth edge sticking up a bit in a slight 'bow'; then evenly tape down the fourth row to create a slight 'bubble' in the paper. But do so gently and neatly, and note as you tape the last edge that you can effect a workman's job of 'smoothing' by gradulism what would have been one big wrinkle into a series of tiny ones; now imagine adding a few stiffening devices behind that 'bubble' so as to stablilize the shape into something a bit more graceful. I believe that is why we can see some suggestion of slight distortion in the covering in some places, such as the Darwin photo, but would agree it's subjective and hard to say for sure.
I would expect the aft ends of the stiffeners and the aft row - toward the aft ends, to be the last to finish. I would expect the aft edge of the panel to have rather evenly 'absorbed' surplus metal - a very slight bowing between fasteners - not greatly noticable from a few feet away. The overall effect should be rather clean to the eye. This should give a very stiff membrane with reasonable restoration of strength and rigidity.
The reasons I'd choose aft over the top, forward or bottom edges to take the last fasteners and to absorb the distortion are threefold -
1) weather dictates a tight upper joint,
2) slipstream and weather dictate a tight forward joint, and
3) the bottom row is more complex since it has a double row of staggered rivet placement, so a bit harder to work the surplus into that gracefully.
The nature of the curves also suggests that the forward or aft edges would be the most logical to absorb the effects of trying to meet the longitudenal curve; the upper and lower edges would not be as useful for that in my view. Aft row is the answer in my view.
I appreciate your question, and of course my 'would dos' are conjecture in this sense - I wasn't there, and I'm doing the best I can to offer some insight into the possibilities that we're looking at where this part was concerned. It could obviously be any number of things, but the part as studied suggests some handling along these lines, as does the use of stiffeners - if it is all in fact related to the Earhart Electra's lav window.
Maybe it can also be seen why some of us are so keen on further metallurgical examination of 2-2-V-1: we may yet glean more details from her damaged state as to how she was installed, where the failure forces eminated and how they translated the full membrane in the various modes of failure that are suggested along her edges. If it leads to the Electra, nice day; if not, we'll have learned a great deal about analysis of this sort.