The 36 gunas, one koota at a time

The 36 gunas are eight separate lookups weighted one point through eight points, computed from nothing but the two Moon positions, and four of the eight pay all or nothing. A total of 22 can be assembled in dozens of ways, and the ways are not equivalent. So the eight are worth taking one at a time: what each looks at, what it is worth, and how often each result actually occurs.

The rates below are counted over every possible pairing. Twenty-seven nakshatras of four padas each gives 108 positions, and every ordered pairing of them is 11,664 combinations, all of which we ran under the same tables. Nothing here is sampled or extrapolated.

Where does the eight-part structure come from?

The koota tables in the form used today appear in the vivaha chapters of the medieval muhurta compendia — works like the Muhurta Chintamani, and set out again in the Jataka Parijata. Modern handbooks print them from those, and calculators print them from the handbooks. Reproduced that consistently for that long, the list has come to read as a single object.

Which particular verse first fixed all eight together, with those particular weights, is not something this page can settle honestly. The components are demonstrably older than the assembly: nakshatra classification into deva, manushya and rakshasa is in the earlier literature, and the animal symbols are older still. Pingree’s work on the layered origins of the Indian astrological corpus is a useful corrective to the idea that any of this arrived in one piece.

So the honest description is a compilation, standardised in print, not a revelation with a single citable line. That is not a criticism. It is the reason two respected sources can carry slightly different tables for the same koota without either being wrong: they are downstream of different manuscripts and different editors. The authority of the number in your report rests on the authority of a table, and tables have editions.

What does the calculation actually take as input?

Two Moon longitudes, and nothing else. This surprises people who have watched a full kundli being drawn and assumed the whole chart feeds the score. It does not. Every one of the eight tests is derived from where each Moon sits.

The chain is short. A birth date, time and place give a moment. That moment gives the Moon’s ecliptic longitude, which is the part that comes from an ephemeris and is settled to a fraction of an arcsecond against JPL Horizons. Subtract the ayanamsa to get the sidereal longitude. Divide the 360 degrees into 27 equal arcs of 13°20′ and you have the nakshatra; quarter each arc and you have the pada. Divide the same 360 into twelve and you have the rashi.

From those two derived labels, everything follows. Varna, vashya and bhakoot read the rashis. Tara, yoni, gana and nadi read the nakshatras. Graha maitri reads the lords of the two rashis. That is the whole input surface: two positions on one circle.

It has a practical consequence. Neither person’s Sun, ascendant or Mars is in the 36 at all, which is exactly why manglik has to be raised separately and why two reports can agree perfectly on the gunas and disagree completely on whether there is an objection. Different inputs, different question.

It also means the score is robust to birth-time error in a way people do not expect. The Moon covers about thirteen degrees a day, so it stays inside one nakshatra for roughly twenty-two hours and one pada for about five and a half. An hour of uncertainty in a recorded time usually changes nothing. A wrong date, or a time zone applied from the wrong historical rule, changes everything.

The four light kootas, and how often they are won

Varna, vashya, tara and yoni carry ten of the 36 points between them, and three of the four are won far more often than they are lost. They are the floor almost every total is built on.

Varna sorts each Moon sign into one of four groups and asks how the groom’s compares with the bride’s. It is worth 1 point and 62.5% of pairings take it. The arithmetic is checkable in your head: four groups spread evenly over twelve signs gives sixteen ordered combinations, and ten of those sixteen satisfy the condition. The measured rate lands exactly there. Our varna koota tool shows which table it used, because the printed editions do not all agree.

Vashya groups the twelve rashis — human, quadruped, water, insect, wild — and asks how the two sit together. Worth 2 points, and in the common convention it is binary: 22.2% of pairings take both points and the rest take none. Sources disagree about where Scorpio and Leo belong, which is one of the quieter reasons two totals differ by two.

Tara counts the distance from one nakshatra to the other and reads it against a nine-fold cycle. Worth 3 points, with 33.3% of pairings taking all three. Its distinguishing feature is at the bottom: 0% of pairings score zero on it. Tara never pays nothing. Its measured average is 2 points. The tara koota tool asks you which counting direction to use, because that choice is real.

Yoni compares the animal symbols the two nakshatras carry against a matrix running from identical, through neutral, to hostile. Worth 4 points. Full marks require the same symbol on both sides, and that happens in only 7.3% of pairings, while an outright zero happens in 7.1%. Almost everybody lands in the graded middle.

Losing yoni points is the normal case. More than eleven pairings in twelve do.

A report flagging a partial yoni score as a concern is flagging the condition of the overwhelming majority. The yoni koota tool prints both matrices in circulation so you can see where they part company.

Why does the pada matter if the nakshatra is the same?

Krittika is the standard illustration. Its first pada falls in Mesha and its remaining three fall in Vrishabha, so two people born under Krittika, days apart, can hold different Moon signs. Half the eight tests read the Moon sign and not the star, which is why the population of pairings is 108 squared and why this particular piece of machinery catches people out constantly.

Twenty-seven nakshatras do not divide evenly into twelve signs. Each nakshatra spans 13°20′ and each sign spans 30°, so a nakshatra can straddle a boundary. That is the whole of the mechanism.

Follow it through the eight tests and the split is clean. Varna, vashya and bhakoot read the rashi. Graha maitri reads the lords of the two rashis, so it reads the rashi too. Tara, yoni, gana and nadi read the nakshatra and do not care about the pada at all. A shift of one pada can therefore move ten of the 36 points and leave the other twenty-six untouched.

Which produces a specific, common and entirely avoidable argument. Two reports agreeing that both people are Krittika, and disagreeing by seven points, are usually not disagreeing about tables. They are disagreeing about which side of a sign boundary a Moon fell on, which is a question about the birth time and the ayanamsa instead of a question about the tradition. That is a far more tractable disagreement than it looks, because it has a factual answer.

It also explains a rule of thumb. If a nakshatra is close to its start or end, the pada is the thing to verify first. The Moon crosses a pada boundary roughly every five and a half hours, so a birth time recorded to the nearest hour is usually safe and a birth time recorded as “early morning” is not.

Twenty-six of the 36 points sit in four tables

Graha maitri, gana, bhakoot and nadi carry those twenty-six points, and the top two of them are decided by a single yes or no. This is where totals are actually made and lost.

Graha maitri takes the lords of the two Moon signs and asks how those planets stand to each other: friends, neutrals or enemies. Worth 5 points, with 38.9% of pairings taking the lot and only 5.6% taking none. Two documented scoring tables are in use and they disagree precisely about the asymmetric cases, where Mercury counts Venus a friend and Venus returns neutrality. The graha maitri tool names which table produced its answer.

Gana sorts the 27 nakshatras into deva, manushya and rakshasa classes and compares them. Worth 6 points. 33.3% of pairings score full and 22.2% score nothing, which is what a crossing between the two classes the tradition treats as opposed produces. Everything else lands in between.

Bhakoot measures the distance between the two Moon signs and rejects certain distances outright. Worth 7 points, and it is a straight yes or no: six of the twelve possible distances are treated as unfavourable, so 50% of all pairings score zero.

Half of the equally weighted possible pairings lose seven points.

Nadi assigns each nakshatra to adi, madhya or antya, and the classical sources treat a match inside the same group as the gravest objection in the system. Worth 8 points, also all or nothing. Three roughly equal groups give roughly the arithmetic you would expect: 33.3% of pairings share a nadi. The nadi page covers what the schools do about it and which cancellation conditions each one records.

Why are the point values one through eight?

Because they are literally the first eight whole numbers, and those add to 36. One, two, three, four, five, six, seven, eight. The weights are not measured magnitudes of anything; they are a ranking, written down in order, and the famous total of 36 is a consequence of that ranking instead of a target anyone chose.

This is the part of the system most open to argument, and the argument cuts both ways. In its favour: an ordinal ranking is an honest way to express “nadi matters more than varna” without pretending to a precision nobody has. Against it: as soon as those ranks are added together and compared with a threshold, they are being treated as though the gap between nadi and bhakoot were the same size as the gap between varna and vashya. One point.

In an ordinal scheme that gap means nothing of the sort.

The effect is visible in the rates. Varna is won by roughly five pairings in eight and costs a single point when lost. Bhakoot is lost by half of possible pairings and costs seven. Adding those two into one number puts a common, cheap event and a common, expensive one on the same footing, then invites a family to read the sum as though it were a temperature. Which is the case for reading the eight components one at a time, and it is a case the tradition itself supports: the classical discussions weigh specific objections by name, not by arithmetic. The distribution of totals shows what happens to the curve when four binary tests dominate a scale that looks continuous.

Four of the eight pay all or nothing

Varna, vashya, bhakoot and nadi. Measured across every pairing, each of those four returns either its full value or zero and never anything in between. The other four — tara, yoni, graha maitri and gana — return graded partial scores.

The split does not follow weight, which is what makes it interesting. The lightest koota and the heaviest are both binary. The consequence is structural: 15 of the 36 points are decided by two coin-flip-shaped questions, so totals cluster around four outcomes with the graded kootas providing the scatter around each cluster. That is why the mean across all pairings sits at 20.1 and why the scale behaves less like a thermometer than like four bands with fuzzy edges.

It also changes what a one-point difference is evidence of. Between two reports differing by a point, no heavy koota has changed hands; a small table has. Between two reports differing by seven or eight, a whole verdict has flipped, and establishing which one is the first thing to do.

What is missing from the list of eight?

More than most reports admit, and the omissions are where refusals usually come from. The 36 gunas are one instrument. They are not the whole of what an astrologer may raise.

Manglik is the big one and it is entirely outside the score, judged from Mars against a set of houses counted from a reference point. Rajju and vedha are treated by several schools as hard vetoes that override a satisfactory total instead of adjusting it. Papasamyam weighs malefic placements on a separate balance. And the southern porutham systems are a different instrument again, scoring out of ten rather than 36, with their own acceptance conventions.

So a report showing a comfortable Ashtakoota total and a refusal is not contradicting itself. It is reporting two instruments, one of which the reader has been trained to look at and the other of which does the actual work. Asking which instrument produced the objection is a fair and specific question.

How would you check one of these by hand?

Start with varna, because it is the shortest chain from birth data to a defensible answer. You need each Moon sign, which is one lookup, and one four-way grouping table, which is printed in every source listed at the foot of this page.

Get the two rashis. Assign each to its varna group. Compare them under the rule your source states. That is the whole test, and it either agrees with the report in front of you or it does not. If it disagrees, the cause is almost always the table and not the arithmetic — which is precisely the useful finding, because it tells you the report is working from an edition you can now go and name.

Bhakoot is nearly as short: count the distance between the two rashis in both directions and check the pair against the list of rejected distances. Nadi needs only the two nakshatras and one three-way table. Between them those three tests account for sixteen of the 36 points, and none of them requires anything more than counting.

Do the three of them and something useful happens even when they all agree. You stop holding the score as an oracle and start holding it as a procedure with steps you have personally walked. A family that has watched one koota being counted out on their fingers argues about the result quite differently from a family that received a printout, and it argues in a register the tradition itself would recognise, because the tables were written down precisely so that they could be followed by anyone who could read them.

The tests that resist hand-checking are graha maitri, where the friendship table has genuine variants, and yoni, where the matrices differ. For those, the thing to ask for is not the arithmetic but the source. A total that cannot be traced back to a printed page is a total resting on somebody remembering a table correctly.