20 spokes marked for inside and 20 for outside and the head angles are slightly different
The first photograph is a spoke from my Vincent next to one I bought from Ron Kemp in the 1990s, showing the latter is bent at 90° but the former at 22.5° less than 90° (i.e. 67.5°).
The Vincent spoke is one from the inside of the spoke flange whose head faces the drum. The second photograph shows the flange, with the camera aligned with the edge of one of the drums to make it clear the spoke flange isn't flat, but rather is bent inwards by some as-yet-to-be-determined angle.
Since the spokes in question seat against the outer surfaces of the flange, the relevant distance is shown in the above to be 3.34". The length of the spokes where they enter the rim is 8½" so the angle the spokes make is sin–1(1.67"/8.5")= 11.3°.
For maximum strength the entire circumference of a spoke head should make contact with the spoke flange, which wouldn't be the case for the Vincent spoke shown in the first photograph
unless it turns out the angle at which the flange is bent inwards is 11°. No matter what, the 90° bend of the Ron Kemp spoke, plus the "missing metal" from part of the slot (shown in green), means relatively little of the metal from the spoke would make contact.
Unless the angle(s) of the spokes matches that of the flange, there will be remarkably little metal in contact. If the angle of the flange does turn out to be, say, 11° it means the 20 spokes whose heads point inwards (i.e. the ones that are presently trapped in my hub) should be bent by 11° more than 90°, i.e 101°, to make full contact.
Anyway, there are two issues the ideal set of spokes would solve. The diameter of their heads would be small enough to slip between the flange and drum, and they would come bent with different angles for ones whose head face in and that face out.