What "dosha" meant before it meant "defect"

The word being translated as “defect” described something to be weighed against other factors, and English handed it a permanence it never carried. The sentence that ends most of these conversations is short. There is a dosha. By the time it reaches an English-speaking family it has usually been rendered as defect, or flaw, or sometimes affliction, and each of those words does something the original did not. They describe a fixed property of a thing. They close the matter.

What did the word carry before it arrived in English?

Fault, yes — but a fault identified in order to be weighed, in a particular context, against merits. The harsh sense is genuinely present in the lexicons and it would be dishonest to pretend otherwise. What is missing from the English is the habit of mind the word travelled inside.

Open Monier-Williams at doṣa and the entry runs long: fault, vice, deficiency, want, blame, accusation, sin, offence, morbid element, disease. That is not one meaning with shadings. It is a word doing work in law, in medicine, in grammar and in criticism, and taking a different specific sense in each. Apte’s Practical Sanskrit-English Dictionary gives a similar spread and orders it differently. A word with that range is a word whose sense is fixed by its discipline, and the discipline here is not the one an English reader will assume.

Its most familiar technical use is in Ayurveda, where the three doshas — vata, pitta, kapha — are not diseases. They are constituents that everybody has, all three, all the time. The clinical literature is explicit about this; a standardisation study in the journal Ayu states that no constituent is ever totally absent and builds its constitutional types out of which dosha predominates, not which is present. Nobody is cured of pitta. The concern is proportion: whether one runs high relative to the others in this person, in this season, at this age. A dosha in that sense is a reading on a dial, not a mark on a record.

The medical usage is a useful anchor precisely because it is so far from a verdict. Whatever one makes of Ayurveda as a system — the American National Center for Complementary and Integrative Health, cited below, is careful to treat it as a tradition under study rather than an established therapy — the grammar of the word inside it is unmistakable. It is a quantity, and quantities are compared.

Why does guna, the other half of the pair, disappear from the conversation?

Because a report has room for findings and no room for a balance, and guna is the side of the scale that only exists when there is a scale. The counterpart of doṣa is guna — merit, quality, virtue — and the two are a pair in the way debit and credit are a pair.

In classical poetics the gunas and doshas of a composition are assessed together, and the treatises are explicit that some faults are occasional rather than absolute: a blemish in one context and not in another, depending on the sentiment being built. That qualification is not a modern liberalisation. It is written into the definition. A critic who listed the doshas of a verse and stopped there would have been doing half the job by the standards of their own discipline.

The pairing survives, unnoticed, in the name of the procedure itself. Guna Milan is the matching of gunas. The thirty-six points are merits, counted up. The system is named after the credits and is discussed almost entirely in terms of the debits — which is a strange thing for a system to do, and a reader who notices it has noticed something real. Look at what the base rates page reports and the asymmetry is obvious: every koota has a full-marks rate as well as a zero rate, and only one of the two ever gets said out loud.

The word did not change meaning; it changed grammar, from adjective to verdict.

How did a weighing become a verdict?

Three things happened, none of them conspiratorial, and each one removed a place where a qualification used to live. The order matters, because each step made the next one easier.

The first is ordinary translation loss. There is no single English word meaning “a factor weighing against, to be considered alongside factors weighing for”. Translators reached for defect because it is short, and the connotation of permanence came along for free. A defect is a property of an object. It does not vary with who is asking or what else is true.

The second is the move from consultation to output. A practitioner reading a chart in conversation could qualify, hedge, weigh one finding against another, and change the emphasis on being told something about the couple. A printed page cannot do any of that. A field on a report either contains a dosha name or it does not, and the reader receives a fact rather than an opinion in progress. The information that was lost is not the finding; it is the tone the finding was delivered in, which was carrying a great deal.

The third follows from the second and is where most present-day damage is done. Software wants a boolean. Something that is a matter of degree becomes a flag, the flag is either raised or not, and once the flag exists it becomes the thing the report is about. This is not an accusation against anyone who writes matching software; it is what happens when a schema has to be designed. A column typed as boolean cannot store “present but light, and here is why I am not weighting it”.

An honest complication, since this account is otherwise too tidy: the narrowing is not purely a modern or a colonial artefact. The classical texts themselves sometimes state findings flatly, and the electional literature is full of unqualified prohibitions. What is modern is the combination of a flat statement with an English word that forecloses, delivered on a page that cannot be argued with, to a family who has no access to the reasoning.

What do the measured base rates do to the word “defect”?

They settle the question arithmetically, because a defect ought to be unusual and these are not. If nearly everyone has it, the word is describing the ordinary case and doing it pejoratively.

There are exactly one hundred and eight nakshatra-pada positions, which makes 11,664 ordered pairings — the whole population of possible matches rather than a sample of it, with no seed and nothing that would come out differently on a second run. We scored every one of them under each koota. Bhakoot returns zero for 50% of pairings, which is exactly half. Nadi returns zero for 33.3%, and nadi is the finding most often treated as fatal. Gana returns zero for 22.2%.

None of those is a surprise once the structure is visible, and none of them is a discovery about couples. There are three nadis holding nine stars each, so two people drawn at random share one about a third of the time. Bhakoot turns on the distance between two rashis and six of the twelve possible distances are the disfavoured ones. These are properties of how the categories were drawn. The full accounts are in the nadi explainer and the bhakoot one, and the gana piece makes the parallel argument about Rakshasa as a classification rather than a character.

The same arithmetic sets the scale everything else is read against. The mean Ashtakoota total over all pairings is 20.1 out of thirty-six, and 66.7% of pairings reach the conventional eighteen-point line. A score that sounds mediocre in isolation is frequently an ordinary score. Attach the word defect to a condition a third of the population shares and the word is carrying considerably more weight than the finding underneath it can support.

Why do only three of the eight kootas produce something called a dosha?

Because naming is doing work that scoring is not, and the three that got names are the three that end conversations. All eight kootas can return zero. Only gana, bhakoot and nadi have a familiar dosha attached to that zero in ordinary speech.

Score a pairing and you will find zeros scattered across the board. Varna returns zero often; 37.5% of all pairings, in fact, which is a larger share than nadi. Tara and yoni both produce plenty of them — yoni awards full marks to only 7.3% of pairings, so the interesting part of that scale is almost entirely below the top. Nobody has ever had a wedding stopped by varna koota. A zero there is a missing point, quietly absorbed into a total, and the family is never told it happened.

What separates the three that acquired names is not their arithmetic. It is that a name turns a number into an entity. “You lost eight points on nadi” is a statement about a scale, and a scale invites the question of what else is on it. “You have nadi dosha” is a statement about the couple, and it invites nothing. The grammatical shift is the same one that happened to the word itself, running one level down: from a measurement to a possession.

The southern vocabulary is instructive here because it went a different way with the same material. The ten poruthams are named for agreement — a porutham is a fit, a suitability — and the negative case is usually stated as an agreement that was not obtained rather than as a defect that was found. That is a real difference in register, not a translation quirk, and it survives even where the underlying test is close kin to a northern koota. It is not uniformly gentler; rajju and vedha do get named and can refuse a match on their own. But the default framing is a scale with a bottom rather than a condition with a carrier, and it is worth noticing that the same astronomy supports both ways of speaking.

None of which argues that the three named doshas should be renamed or ignored. It argues something narrower: that the difference in how they land on a family is partly a fact about vocabulary rather than entirely a fact about charts, and that this is checkable by looking at the scores the other five kootas quietly produce every day.

Which half of the classical passage gets left out?

The half immediately after the finding, where the same author sets out the conditions under which the finding is held not to apply. The exception is part of the rule, written by the same hand, often in the same sentence.

The classical treatments that name these doshas rarely stop at naming them. The parihara literature — the material on setting a finding aside — runs alongside the material that produces findings, and the muhurta tradition is where it is thickest. Look at the structure of a work like the Muhurta Chintamani, or at how the Brihat Parashara Hora Shastra handles its own prohibitions, and the pattern is plain: a condition is stated, then the circumstances that neutralise it. Nadi dosha is described as not applying where certain conditions of rashi or lord are met. Comparable exceptions attach to most of the named doshas.

A report that prints the finding and omits the exceptions has not summarised the tradition. It has quoted half a sentence, and it is the half that turns a weighing into a verdict — the same narrowing the word itself underwent, performed again at the level of the document. This is why the engine behind these pages cites the text each rule comes from and reports the cancellations that were tested and did not fire alongside the ones that did. A cancellation checked and not met is information; a cancellation never checked is a silence that reads like a finding. The cancellation reference and the matching-veto reference keep the two opposite rule types separate.

Is the harsh reading simply correct, and this whole argument sentimental?

It is the strongest objection to this piece and it deserves a straight answer rather than a dismissal. Someone can reasonably say: the lexicons list sin and blame among the senses, the texts do state prohibitions flatly, and softening a tradition to make a stranger feel better is its own kind of dishonesty.

Two responses, and neither of them is that the tradition is wrong. The first is that the claim here is narrow. It is not that doṣa once meant something benign. It is that its use across the disciplines was comparative and contextual in a way English “defect” is not, and that a translation which drops the comparative frame is a worse translation even by a strictly traditional standard.

The second is that the flat prohibitions and the cancellations sit in the same corpus, often the same chapter. A reader who wants to be strict about the tradition has to be strict about both halves. Raman’s Muhurtha is a convenient example in English because it states its rules firmly and then states its exceptions just as firmly, with attributions. Quoting only the firm part is not rigour. It is selection.

Where is this account uncertain?

In two places, and both are worth stating rather than smoothing over.Etymological stories are easy to tell and hard to prove, and this one is being told to make a point, which is exactly the condition under which such stories become unreliable.

The first uncertainty is about how much weight the semantic history can bear. Words shift. Sanskrit technical vocabularies are not one vocabulary, and the fact that doṣafunctions as a quantity in Ayurveda does not automatically mean it functioned as one in jyotisha. The Ayurvedic case is presented here as an illustration of the word’s range, not as proof about a different discipline.

The second is that the parihara literature is not uniform. Traditions disagree about which exceptions are canonical and which are late additions, and a cancellation accepted in one lineage is rejected in another. At least one exception that practically every practitioner can recite has no locatable printed source in the editions we have checked, and it is reported as traditional and unverified wherever it applies rather than presented as a citation. A rule with no text behind it is not thereby false; it is simply not the same kind of thing as a rule with a page number.

What should you ask when the word arrives?

Three questions, and every one of them is a request for detail rather than a challenge. None requires you to know anything about the subject beforehand, and an astrologer working from a stated method will find all three ordinary.

Which koota, or which planet, produced it — by name? A label alone cannot be compared with anything. A named rule can be checked against any other practitioner and any printed source.

What was the score, out of what maximum? Nadi carries 8 of the thirty-six points and varna carries a single one. “There is a dosha” conceals which of those two just happened, and the difference between them is most of the total.

And which cancellation conditions were tested, with what result — including the ones tested and not met? That last clause is the one that does the work, because it converts a silence into an answer.

A finding that survives all three is a finding worth taking seriously, and some do. Most of what reaches a family as there is a dosha has never been asked any of them, and the word has been doing the arguing on its own.