the third pic that phil has posted is the right one...
There are various configs as phil said...
as it can be seen the metallic plate has two different orientations...
the same applies for the plastic plate where the metallic is mounted onto...so there are 4? different orientations? cant do the maths now
as it can be seen in phils last pic the plastic frame is different in each side...one of the sides is wider than the other.That side if it is the wider has to be mounted with the thinner side of the metallic frame...to lets say "balance" it or center the evaporator

....
The mach 2 evaporator can be mounted correctly in two ways...
first of all the evaporator can be mounted lets say with the hypothetical line coming through the two evap head thread holes horizontally....and the second one is with the hypothetical line coming through the two evap head thread holes vertically BY TURNING THE EVAP CLOCKWISE by 90 degrees...
THESE ARE THE ONLY TWO CORRECT WAYS to mount the evap head...this is coming directly by nventiv and to be more specific it has been mentioned many times by JCViggen known in XS forums who was working for nventiv and who achieved the 5ghz overclock from the 3.2ghz prescott in ce bit when the mach 2 GT was launched-introduced...
so only these two evap configs are correct and will flood the evaporators labyrinth correctly without stressing the capillary tube and making damage...
I have tried with my last mach 2 mounting different the evap tube by turning it anticlockwise...i had no probs with the mach 2 and i experienced the same performance too...but still it may be risky mounting incorrectly

...
phils last pic is the correct one even though it shows that the socket covered isnt centered by the evaporator...