Hi Martin,
I am perfectly happy to grant that some Indian texts, including the YJ, make use of a tropical year. (I haven't got the Sanskrit text to hand now, but I presume Pingree's translation to be correct.) The point is that the calendar and the astrological use of the zodiac are two different things. Insisting that they must be the same is a tropicalist prejudice which, if applied to the Indian texts, will certainly make them confusing.
I am afraid, you are wrong. Where do ancient Indian texts state that calendar making and astrology have to be separated? The Yavanajataka is an astrological work from its beginning to its end. The last chapter 79 is astronomical and contains instructions how to calculate the planets and all the other factors required for an astrological chart. This is the context in which the tropical year length appears, and it means that tropical signs were used in practice.
Besides, the chapter starts with the following verse (translation by Pingree):
?The wise say that the observed course of the planets is the supreme eye of the entire body of the rules of horoscopy (sarvasya [hor?]vidhisa?grahasya, D.K.). I shall explain it concisely according to the instruction of the Greeks.?
(Note: hor? is a conjecture by Pingree to make the verse complete, but it is obvious that vidhisa?graha refers to the whole astrological teachings of the book.)
This means: What follows in ch. 79 serves astrology, not calendarmaking!
Honestly, I wonder a bit how seriously you looked into the text.
Also note that the Suryasiddhanta was the basis of astronomy, calendar, and astrology. There was no such fundamental difference and separation between these things.
The point is that the text does not define the zodiac as beginning with the equinox, etc; rather, it locates these points within the (apparently sidereal) zodiac. For the meaning you wish the passage to have, a congruence in case would be the normal construction in Sanskrit. (For example, *raver uda?m?rgagati? m?g?di? ... vidy?t)
Well, you say "normal construction in Sanskrit", but imho such judgments are dangerous because, as no doubt you know, Sanskrit is extremely different depending on literary genre, tradition, and epoch of the text. The correct approach to settle the question is to study the usage of language in ancient astronomical/astrological texts. If you do so, it will turn out, in my opinion, that you are wrong.
The question concerns the exact meaning of the word -?di in compounds preceded by the name of a zodiac sign in astronomical texts. And whether a locative case is used or not does not really matter. Even with other declension cases one could discuss whether ?di means the exact beginning or the first couple of degrees.
Let us look again at the verse I quoted from the Suryasiddhanta (14.9):
bh?normakarasa?kr?nte? ?a?m?s? uttar?ya?am
karky?destu tathaiva sy?t ?a?m?s? dak?i??yanam
?Six months from the ingress of the Sun into Capricorn are the northward course (of the Sun),
and six months from the beginning of Cancer are the southward course.?
The first line of the verse states that the uttarayana months start at the ingress into Capricorn. The second line has the dakshinayana beginn from the "beginning of Cancer". It is obvious in this case, that "beginning of" (-?di-) does *not* mean "somewhere in the first degrees of Cancer", but it means the *ingress*.
Another verse from Aryabhatiya (4.1):
me??de? kany?nta? sama? udagapama??al?rdham apay?tam
tauly?de? m?n?nta? ?e??rdha? dak?i?ena eva
"The first half of the zodiac that begins with Aries and ends with Virgo has northern declination. The other half that begins with Libra and ends with Pisces has southern (declination)."
This statement is only true if the equinoxes are exactly at the beginning of Aries and Libra and if "begin" (?di- again) refers to the exact point of the beginning of the sign.
Another verse from Brhatsamhita (3.1) (referring to Vedangajyotisha):
??le??rdh?d dak?i?am uttaram ayana? raver dhani??h?dyam
n?na? kad?cid ?s?d yenokta? p?rva??stre?u
"In the middle of the lunar mansion ??lesh? (or at the end of sidereal Cancer) started the southward path of the Sun, and at the beginning of the lunar mansion Dhani??h? (or at the end of sidereal Capricorn) started its northward path / in ancient times. This is stated by old scientific works."
To my knowledge, no expert ever understood this verse in such a way that "middle" (ardha) and "beginning" (?di) mean only "within a few degrees". All assume that accuracy is intended.
Please show me texts, where ?di means "somewhere in the first degrees of".
Conclusion: It is quite obvious that the Yavanajataka in the verse under discussion puts the equinoxes at the exact beginning of the signs. Even more so when its context is taken into account: ephemeris algorithms based on the tropical year.
Dieter