Fastenings are not so difficult. Most won't be far off the 3.5 / 5 ratio, though you'd probably want to use screws with size-smaller heads in many areas for appearance or accessibility – a spanner takes up a lot of room!
A rough guide to the right area is to assess the nut and bolt sizes in full size, and bear in mind they had to be manipulated by blokes with spanners. Looking at preserved machinery, a lot of the ordinary fastenings were no more than perhaps 3/4 diameter; much above an inch was usually for major, fitted bolts holding parts like cylinders to frames; or specials on marine-engine frames. The designers calculated joint strengths based on several moderately-sized rather a few huge, bolt or rivet shanks – and also considered how the thing was to be erected and serviced. .
The example Paul cites of a 1/4" bolt in 5" scale(?) creating a theoretical 0.175" bolt, equates practically to 2BA (0.180" dia measured from a sample screw), possibly M5 though M4 might be better for appearance.
I think we mean gauge here. The difference between scale and gauge?
A 1/4" diameter bolt on 5"-scale drawings is 0.6" dia in full size (5/8" in practice).
A 1/4" bolt on a 5" gauge loco (approx. 11:1 scale) would be nearly 3" diameter in full size!
'
The easiest way to re-scale the drawings is to use a spread-sheet, because you need only drag-copy the formula down the column.
Round plate thicknesses and rod diameters sizes to 1, maybe 2, decimal places; to cope with only metric materials being available create a further column converting them to mm ( X 25.4). For plate and sheet work it is probably better to go up a size but don't over-do it.
NB – Having to change the plate thickness from the original plans then compromise for stock sizes, can affect a lot of other dimensions, especially so in locomotive frames and traction-engine horn-plates with many components on both sides of the plates.