The discussion alluded to above in Wijesekera (Buddhist and Vedic
Studies) is quite brilliant and needs to be consulted before coming
to any conclusions on this very interesting question. Among his
points:
The etymology deriving gandhabba from gandha is folk etymology
('ludicrous'), and mentions of the gandhabbas as associated with
scent (which happens from the Vedas on) are simply punning on the
similar sounds. (Gantabba is a similar folk etymology). In fact the
etymology is obscure.
The term is thouroughly embedded in the Vedic mythology and
virtually all the aspects of gandhabba/gandharva in Pali can be
traced to Vedic antecedents.
The term is certainly cognate with the Avestan (Persian) gandarewa,
and probably cognate with the Greek kentauros (centaur); thus it is
certainly Indo-Iranian and probably Indo-European.
Its oldest mthic connotation is of a spirit of the water; a monster
of the watery abyss; horselike; also associated with the clouds and
the sun. Since the rain is held to be the male fertilizing the
Mother Earth, the Gandharva becomes associated with fertility, and
is represented as similar to randy satyrs (cf Pancasikha). There is
one of the recognized forms of marriage in Hindu culture called
the 'Gandharva' wedding, which is essentially just elopement. It
also becomes associated with the spirits of the dead, and is
associated with the process of rebirth.
This is an all too brief summary of an extensive survey, raising
many interesting points about the way the Buddha used the ideas
current in his culture. It is apparent that in such cases, where the
Buddha was clearly invoking a Vedic mythology rich in connotations
(incidentally, as Wijesekera points out, the passage on the 'coming
together' of the three things is used in the Assalayana Sutta, where
it is put in the mouth of the risis of old, thus explicitly being
presented as a pre-Buddhist teaching), it is in that mythology that
we should seek for the primary context.
The Theravadin commentators were, it seems, not well acquainted with
the Vedas, so they cannot be expected to appreciate this context.
Thus the commentarial statements explain how such references fit in
with the developed Theravadin teaching, but do not tell us what the
statements meant for the original audience.
The mention of the gandhabba here does not, in and of itself, prove
anything regarding the controverted 'in-between state',
(antarabhava), although some of the schools that accepted the
antarabhava did indeed explain the gandharva in this way. I think
the original suttas did accept an antarabhava, but i do not base
this conclusion on the current passage. That being said, the present
passage obviously is consistent with an antarabhava.
It should hardly need repeating that whether there is an antarabhava
or not makes no difference whatsoever to the principle that what
takes rebirth is the samvattanika vinnana, not the vinnana of which
Sati says is 'tad'eva vi~n~naana.m ana~n~na.m'
There is a version of this passage in the Ekottara Agama (EA
12.21.3; translated by Bhikkhu Pasadika in the Buddhist Studies
Review, Vol 20,1. We have permission from Bhikkhu Pasadika to use
these translations and will include them on suttacentral). This
essentially just takes the passage on the conditions for conception
from the Mahatanhasankhaya and Assalayana Suttas and presents it as
a discourse in itself, adding that one should 'cut off' these
conditions. It adds several variations to the conditions mentioned
in Pali, for example, the presence of lust, anticipation of
offspring, etc. The rendering of what is presumably 'gandharva' (or
a dialectical variation) is translated by Pasadika
as 'consciousness "from outside"' or 'consciousness-genius', which
he notes as unique to EA. Anyway, it seems that the EA tradition
accepts the gandharva as a kind of consciousness. As this was a
controversial point, this presentation might be of use in
determining the dubious sectarian affiliation of EA.