Today being a sunny day here in Rome I went to my balcony with the two incident light meters I own: a Gossen Sixtino II and a Gossen Multisix. They were both set for incident light reading.
With each of them, I did three readings: 1) straight toward the sun; 2) in a direction which was around 50° away from the sun (lateral lighting); 3) pointing the device opposite to the sun (my body projecting a shadow on it).
Case 1): 1/125 @ f/11 + 0,6EV for both devices. Perfect concordance.
Case 2): 1/125 @ f/8 the Multisix, 1/125 @ f/4 the Sixtino;
Case 3): 1/125 @ f/4 both instruments, perfect concordance.
I draw some conclusions from this test:
-) The Sixtino is "useless" in incident light unless pointed toward the sun (light source). The sliding thing in front of the cell does not intercept light falling on it from the sides. It works more like a disk than like a sphere, this is not explained in the user manual.
-) Using the Sixtino as per instruction manual is in a few words a waste of film; That's something I noticed as soon as I bought it. The "incident" light reading was more of an "indecent" light reading. But using it to base exposure as 2F/2F suggests works and gives results that are consistent with what I would obtain with any reflective light meter (or the sunny 16 rule if you prefer).
-) The Multisix gives 1,6 EV of difference in the "canonical" use and the "un-canonical" use. That's way too much to be ignored when using slides and probably also when using colour negatives. The sphere of this light meter does not "equalize" the light coming from 2/3 and the light coming from the front. The reading is different and so the direction of use is determinant in the reading. This was not obvious to me before this test. I thought that the task of the sphere was to internally bounce the light so that a lateral light and a frontal light don't give very different readings. But that's not the way it is.
The case may be that other incident light meters have a more "equalising" sphere though.
PS I'll begin using those bloody things in my daily practice for a few rolls, as the use of the spot reflective light-meter is a bit slow and cumbersome at times and incident light metering can definitely be practical for the photography I practice. I'm glad that the Sixtino II can be used for incident reading anyway, provided I ignore the instruction manual. This can be useful info for some other photographer, I hope.
EDIT: I would like to stress that if I were to make a self-portrait of myself in position 2, using the Multisix as indicated by the common wisdom would very likely lead to overexposure of the sunny side of my face. Using the Multisix as suggested by 2F/2F would lead to a very predictable 1/125 @ f/14 or so, which would likely place the sunny parts and the shadowy side of the portrait just right on the film. I encourage anyone to perform the same test and report experiences and considerations.
EDIT II: valued above obtained with light meters set at 100 ISO, just for clarity.