In my opinion, many of the drawings available on the interweb are inaccurate in one or more respects. I recently downloaded a sample vector drawing from one "Blueprint" site to see if it was compatible with my Illustrator software. Even though it was a good looking, detailed piece of work, I discovered that elements of the side and top view were differently placed longitudinally. That's not a confidence builder.
One of the disappointments encountered in creating, and I use the word deliberately, the top or front view of a three-view drawing is very precisely tracing one side in Illustrator and then copying and flopping it to provide the other side. Almost invariably, the flopped side does not match the other side of the original. (It was possible in the good old days when drawings were created with T squares, triangles, and French curves to get the two sides the same within a line width by tracing a master but not worth the extra effort.)
In any event, pre CAD, the shape of the airplane was not established by three-view drawings, per se, but by 3-D solid master models created from loft-line drawings produced by engineering that were consistent with the detail part and assembly drawings.
That said, it is possible by using manufacturer's station drawings, dimensioned three views, loft-line drawings (relatively rare, even rarer in a usable size), data (e.g. root and tip chord dimensions and wing-sweep angle at the quarter chord to establish an exact basic planform), maintenance manual illustrations, and photos (which can be misleading due to the various distortions introduced by the camera) to create a pretty good three-view drawing. That takes a lot of time and is unlikely to be perfect but it is the best a good draftsman can do without being able to measure a real airplane, which involves significant effort and still doesn't provide more than an approximation of shape as opposed to size.
As others have noted, the primary purpose of three-view drawings produced by manufacturers was to illustrate significant dimensions, not provide model kit manufacturers with an accurate depiction of the aircraft's shape. My favorite example is this McDonnell F4H-1 top view, which provides the different lengths of the short-nose and long-nose Phantom IIs on the same drawing. (The side view and presumably this top view are the long nose version.)
The major source of confusion is whether the cited length was that parallel to the water lines or parallel to the ground with the aircraft at its normal stance (there's actually a detailed definition of the variables like weight that affect the stance). In this case, the length also includes the barricade-strap deflector (to prevent the strap from hanging up on a gun barrel and jerking the airplane sideways) that extends just forward of the nose itself.
Another problem with three-view drawings like those in Standard Aircraft Characteristics charts is that the creator of the original might not have reduced all of the views by the same proportion. One needs to check the common dimensions on each of the views (the wing span on the top and front views, for example) to verify that they are the same. Another problem is differential "stretch" at some point in the reproduction process, which means that the width of the copy might be proportionally different to the height of the original.
The next level up in terms of achieving accuracy is a station drawing:
It's interesting enough when the aircraft is lengthened so the nose extends forward of station 0 (this means you have to delete the minus sign of a station forward of 0 and add it to a station aft of 0 to get the distance between the two stations) but it can get downright confusing when the manufacturer tries to explain it, as in this note on the Grumman F9F-5 station drawing, explaining the changes due to the plug added in the F9F-2 Panther forward fuselage:
The best source of shape is a high-level loft-line drawing that provides cuts (cross sections):
As previously noted, even three-view drawings produced by the manufacturers can't be trusted without close examination. In the case of the P-80, available Lockheed drawings were clearly inaccurate in some respects, e.g. the side view in the TO-1 SAC had inlets that weren't even close to the real thing as well as a TO-1 drawing with a P-80A canopy instead of the early F-80C canopy that it had. Another disappointment was the fact that even after correcting for slightly different horizontal and vertical sizing of the umpteenth-generation copies, matching key points as precisely as possible, etc. the outlines of three different side-view drawings are not identical.
So the trick is somehow determining whether a given drawing is fairly accurate in shape, basic dimensions, location of major components, etc. so it can be used to evaluate the accuracy of a kit and correct any objectionable differences. If you don't have access to manufacturers drawings, the
primary way to determine whether or not a drawing is to be relied on is
to compare the basic dimensions to a dimensioned drawing and the shape to photos (with the caution that photos can be misleading due to perspective distortion, etc.).