Ashtakavarga

Ashtakavarga scores each of the twelve signs by counting contributions from eight sources. Each of seven grahas gets its own table, the Bhinnashtakavarga, and adding all seven gives the Sarvashtakavarga.

The calculator above computes all of it. It also does something no other Ashtakavarga calculator I could find does: it names which textual tradition the tables came from, and shows you the same chart computed the other way.

That turns out to matter more than it sounds.

How a bindu gets placed

The mechanism is simple and worth stating before anything else.

For each graha, there is a list of houses associated with each of eight contributors: the seven grahas themselves plus the Lagna. A contributor places a bindu, a point, in each of its listed houses counted from where that contributor itself stands.

So if the Sun's table says Venus contributes at houses 6, 7 and 12, and Venus is sitting in Mesha, then bindus go into the 6th, 7th and 12th signs from Mesha: Kanya, Tula and Meena.

Do that for all eight contributors and you have the Sun's Bhinnashtakavarga: twelve numbers, each between 0 and 8. Repeat for the other six grahas. Add the seven tables sign by sign and you have the Sarvashtakavarga.

Conventionally, five or more bindus in a sign reads as strength and two or fewer as weakness, which is why the tool marks those.

The totals, which are the obvious way to check the work

The seven tables have fixed sizes that every source agrees on:

Those numbers are a genuine gift to an implementer. Sixty-odd rows of house lists are exactly the sort of data that gets transcribed wrong, and here you have a hard arithmetic check on every one of them. Get any row wrong and a total breaks.

I used it. My own first draft of the Venus table summed to 51 instead of 52, which is how I knew to go and establish the tables properly rather than typing them from memory.

And the error the totals cannot catch

Here is the part worth the page existing.

The contributor tables come down in two textual traditions:

Parashara, in the Brihat Parashara Hora Sastra, chapter 66.

Varahamihira, in the Brihat Jataka, chapter 13, slokas 1 to 7.

They differ in exactly four rows and six cells:

Now look at what those differences do to the counts.

Moon-from-Moon gains the 9th under Parashara. Moon-from-Mars loses it. The two changes cancel exactly, so the Moon's table totals 49 either way. Moon-from-Jupiter swaps the 2nd for the 12th, same count. Venus-from-Mars swaps the 4th for the 5th, same count.

Every difference between the two traditions is count-preserving. Both recensions total 48, 49, 39, 54, 56, 52 and 39, and both sum to 337.

So the checksum, the one reliable test an implementer has, passes identically for both. A calculator can carry the wrong tradition's tables indefinitely with nothing whatsoever looking out of place.

That is an unusually nasty failure mode. It is not that the error is hard to find; it is that the natural test for exactly this kind of error is structurally blind to it.

How much it changes

Not enormously, and not negligibly.

On the birth the page defaults to, the two recensions give these Sarvashtakavarga rows:

Parashara    34 26 24 23 27 34 23 26 32 31 28 29
Varahamihira 34 27 23 22 27 34 23 27 32 31 28 29

Four of the twelve signs hold a different number of bindus. Both rows total 337.

A four-bindu swing across four signs will not usually reverse a reading, but it can cross the conventional thresholds: a sign on 23 under one tradition and 22 under the other is on one side or the other of nothing in particular, but a sign moving from 5 to 4, or 2 to 3, crosses lines that readings are actually built on.

The tool shows both rows so you can see whether your own chart is affected.

How I established which was which

Worth describing, because the answer was not what I set out to find.

I ran four independent research passes over the tables: classical texts and their translations, modern reference works, open-source implementations read directly from source, and a targeted pass on the rows most often misprinted. Then three adversarial checks: one purely arithmetic, one against sources it had to find for itself, one on internal structure.

The arithmetic check passed cleanly: all seven totals correct, 337 overall, no duplicate houses, all values in range.

The source check refuted the result, and not because the numbers were wrong. They were the wrong author's. The 56 rows matched Varahamihira at every cell and differed from Parashara in exactly the four rows above.

The strongest corroboration came from an unexpected direction. PyJHora, an open-source jyotisha library, carries exactly six configuration switches for exactly these six cells, grouped under a comment naming the Parashara–Varahamihira discrepancy. Someone else had hit the same wall and built the same escape hatch, which is about as good as independent confirmation gets for a question like this.

What this page deliberately does not compute

The reductions. Trikona sodhana and ekadhipatya sodhana reduce the raw bindus before some readings use them. They are a further layer with their own textual disputes, and they are absent here rather than partially applied.

There is a related detail that illustrates why half-implementing is worse than omitting. Varahamihira carries a companion rule that the 10th house from the Moon is effectless in Mars's table, neither benefic nor malefic. Without it, his own worked reduction example does not reproduce. So a calculator using Varahamihira's tables and Parashara's reduction procedure is combining two traditions in a way neither text sanctions, and the output belongs to no source at all.

The Lagna's own Bhinnashtakavarga. It exists in BPHS, but its Moon and Venus rows differ from the Lagna-as-contributor rows used in the seven graha tables. Deriving it by reusing those rows would be wrong, so it is omitted. The Sarvashtakavarga total of 337 is the seven-graha sum in any case.

Why another site gives different bindus

The recension, and this dominates. Four rows, six cells, invisible to the totals.

Whether reductions were applied before the figures were displayed. A reduced Sarvashtakavarga does not total 337, so if a site shows a total other than 337 it has either reduced or erred, and it is worth finding out which.

The birth time. The Lagna is one of the eight contributors, so an error in the ascendant moves bindus in all seven tables simultaneously. This is not a small sensitivity: a lagna one sign out shifts the Lagna's entire contribution by one sign in every table.

The ayanamsa, which can move any graha across a sign boundary and thereby shift that graha's whole contribution by one sign.

Whether the Lagna's own table is included in a "Sarvashtakavarga", which changes the total away from 337.

Why eight, and why these houses

Two questions the tables invite, with honest answers.

Why eight contributors? The name says it: ashta is eight, varga is a division or group. Seven grahas plus the Lagna gives eight, and the Lagna's inclusion is what makes the system depend on the birth time rather than the birth date alone. Rahu and Ketu are excluded throughout, consistent with their exclusion from most positional scoring in the tradition, presumably because they are points rather than bodies and have no house rulership.

Why these particular houses? No source I found derives them. They are given as received doctrine, table by table, without argument.

What can be observed is that the lists are not random. Houses 3, 6, 10 and 11 appear far more often across the 56 rows than 8 and 12 do, which matches the general disposition of those houses elsewhere in the tradition: the upachaya houses are treated as favourable and the dusthanas as difficult. So the tables are at least internally consistent with the rest of the system's house doctrine.

That is a statement about coherence, not about origin. A rule with a stated derivation can be checked against its own reasoning; one received without a derivation can only be checked for consistency and against other copies of itself. Which is precisely why the recension question matters so much here: with no derivation to appeal to, the text is the only authority, and there are two texts.

Reading the tables

Three things the numbers say directly, without interpretation.

A sign with high Sarvashtakavarga has many contributors pointing at it. That is all a high number means arithmetically: more of the eight sources listed a house that happens to land there.

The range is narrower than it looks. Sarvashtakavarga values cluster: with 337 bindus across 12 signs the average is 28, and real charts mostly run between about 22 and 34. A sign on 34 is not four times better than one on 23 in any sense the arithmetic supports; it is about 20% above average against about 20% below.

The Lagna's contribution is one-eighth of everything. It is one contributor among eight in every table, which is worth remembering when a reading leans heavily on Sarvashtakavarga and the birth time is approximate.

What this page will not do

It will not tell you what a strong or weak sign means for your life.

What it computes is a count: for each sign, how many of the eight contributors place a bindu there, under a named textual tradition, from planetary positions you can verify against any ephemeris. Every step is arithmetic you could do by hand with the tables and a pencil, and the tables themselves are printed on the page.

The reason to care about the recension is not scholarly fussiness. It is that this is a case where the standard verification passes for both a right and a wrong answer, and the only defence is for a calculator to say which source it used. Most do not, which means most of the Ashtakavarga figures circulating cannot be traced to a text at all.