Why dosha cancellation rules differ between traditions

A cancellation is a named condition under which a tradition sets aside a dosha it has already found, and written that way it is exactly as examinable as the dosha. The trouble is that the two almost never arrive together. The dosha comes first, stated as a finding, with a koota name and a number attached. The cancellation comes later, from a different person, phrased as reassurance: it gets cancelled, do not worry about it. One of those two statements is checkable and the other, as delivered, is not. That asymmetry is why families end up trusting whichever astrologer they spoke to most recently.

A parihara does not mean the dosha was not there. It means the tradition itself carries a documented exception, and the exception either fits this couple or it does not. Nothing about that is soft, and nothing about it requires taking anyone’s word.

What does a cancellation rule have to carry to be checkable?

Four things: the condition stated precisely enough to test, the text it comes from, how authoritative that text is, and which schools actually apply it. Drop any one and the rule stops being verifiable by anyone other than the person who said it.

Take one of the most commonly invoked. Nadi dosha may be set aside when the same planet lords both janma rashis. That is B.V. Raman, Muhurtha, pages 50 to 51, exception 5(b). It is applied in the South Indian Raman lineage, in the North Indian Charak lineage and by Drik Panchang. It restores the full 8 points that nadi is worth. Every clause in that paragraph can be checked by someone who is not us, which is the only test of a citation that matters.

Compare the same claim in its usual spoken form: nadi dosha gets cancelled if the rashi lords match. What has gone missing is which text, which schools, and how many points move. Without the text, a family cannot tell an orthodox rule from a regional habit. Without the schools, they cannot tell whether the astrologer their relatives will consult next holds it at all — and that is the practical question, because the next conversation is where the trouble usually is.

The third clause deserves more attention than it gets. Sources are not interchangeable, and a serious rule set should say which kind each one is. A classical verse, a modern authority such as Raman or a named panchanga, and a popular rule with a single weak attestation are three different weights of evidence, and a report that renders all three in the same typeface has thrown away information the reader needs. Our cancellation reference shows the source confidence for exactly that reason, and a couple of the popular-tier rules would not survive contact with a sceptical relative — which is useful to know before you cite one at a family meeting.

Why does a cancellation have three states and not two?

Because “no cancellation applies” and “we could not tell whether a cancellation applies” produce the same score and mean entirely different things. Collapsing the two is easy to do and hard to notice, and it makes an engine quietly stricter than the tradition it claims to implement — because a question that was never asked ends up filed as a question that was answered no.

• Fired. The condition was checked and is met.
• Not fired. The condition was checked and is not met.
• Unevaluated. The condition could not be checked, because the data required is not present — no birth time, so no ascendant, so no navamsha lagna to test.

A rule that could not be checked is a question still open, and it is often the exact question worth taking to a second astrologer.

Two of the mangal dosha exceptions are permanently in the third state for honest reasons, and we say so rather than letting them look like failures. One of Desabhatla’s conditions asks for a strong Jupiter or Venus in the lagna or the seventh. Strength there means shadbala, a six-fold strength computation this engine does not perform, so we evaluate the placement only and record that the rule therefore fires more often than its source intends. Another rule from the SKU panchanga requires Mars to be retrograde, and our positional layer returns longitudes rather than motion, so that rule returns unevaluated every time. Neither is reported as not-firing, because that would be a confident false statement in precisely the place this product exists to prevent them.

The birth-time case is the one that will affect you. Several of Raman’s exceptions test navamsha positions, and a navamsha needs an ascendant. The planetary side of that is the settled part: positions come from JPL’s DE440 ephemeris, and where the planets were is not in dispute. The ascendant is the fragile part, because it needs a time and a place resolved against the historical time-zone rules in force at that moment. If the time is missing, or rounded to the nearest hour, those rules are unevaluated. A report printing them as not applying has reported a certainty it does not have. You can see how far your own chart moves under a time shift with the birth-time sensitivity tool, and for some charts the answer is not at all, which is worth knowing too.

Which cancellations do most of the work?

Three doshas account for most refusals, and each carries a cancellation family with sources behind it. They are also the three with the highest base rates, which is not a coincidence: a finding common enough to block a large share of matches attracts exceptions, because the tradition has to live with its own arithmetic.

Nadi, worth 8 points, scores zero for 33.3% of all 11,664 equally weighted possible pairings, and the arithmetic behind that is set out in nadi dosha, one in three. Raman’s exception 5(c) sets it aside when the two rashi lords are friends; 5(a) when Rasi and Rajju kutas both prevail. A separate family of rules, attributed to Narada and carried by the VSKP and SKU panchangas, holds that a common janma rashi with differing nakshatras relieves both nadi and gana.

Bhakoot, worth 7 points, scores zero for 50% of pairings, which is to say exactly half, because six of the twelve rashi distances are counted inauspicious. Raman’s exception 6 cancels it when both rashis are owned by one planet or the two lords are friends, and OP, Drik Panchang, VSKP and RMP state the same. That figure is not an estimate: we ran every one of the 11,664 possible pairings and counted, which is what makes an exception set worth examining separately. Frequency alone does not explain why exceptions arose or show that they improve predictions. The Bhakoot guide separates the table frequency, directional reading and documented exception conditions.

Gana, worth 6 points, has one of the more surprising exceptions: the evil may be ignored if the bride’s asterism lies beyond the fourteenth from the groom’s. That is exception 8, and Raman is explicit about the permission rather than hedging it. Graha maitri, worth 5, gets exception 3 — it need not be considered at all when the janma rashis are one and seven to each other, which is a whole koota switched off by a geometric relationship.

One structural decision worth naming, because it changes what a number means. We do not silently adjust the total. The raw thirty-six-point figure stays, and a second figure with the cancellations applied is reported beside it, with the rules that produced the difference named. Astro-Vision states the reasoning for the strict approach plainly — in a weighted point system, many compensating factors already earn their own points, so applying cancellations on top double-counts them. Both numbers are defensible. Printing one without saying which it is, is not.

What is actually inside the mangal dosha exception set?

A large numbered family, mostly from the Desabhatla slokas and the panchangas, sorting into four recognisable kinds. It is worth seeing the shape of it, because “there are cancellations for manglik” is a much weaker statement than the set actually supports.

The first kind is dignity: Mars in its own sign, Aries or Scorpio; Mars exalted in Capricorn; Mars in its own or exalted sign in the navamsha; Mars vargottama. The Sanskrit is quoted directly in the Desabhatla compilation — svakṣetre, uccāṁśe svāṁśage vā — which is the useful thing about that source: it prints the verse rather than a summary of it.

The second is company: Mars conjunct or aspected by Jupiter, by the Moon, by Mercury; Mars combust within about seventeen degrees of the Sun. The third is position: Jupiter or Venus in the lagna; the Moon in a kendra from the lagna; Mars in a movable sign; Mars at a house junction.

The fourth kind is the interesting one, because it looks at the other person. One rule cancels the dosha when Saturn occupies one of the same five houses in the other chart. Another applies when the groom’s chart carries more malefics in those houses than the bride’s — Desabhatla’s pāpādhikye vare śreṣṭha, which is papasamyam in nucleus. And one ties mangal dosha directly to the guna score: rāśi maitraṁ yadā yāti gaṇaikya vā yadā bhavet — rashi friendship, or matching gana, or simply a high total, and no mangal dosha obtains. That last one is the only classical rule we have found that links the two systems arithmetically, and it deserves to be better known than it is.

Then there is the best-attested exemption in the whole registry, which is not a cancellation at all but a limit on the finding: Raman’s sign list, stated identically by the OP, DP and RMP panchangas. Mars in the second is bad provided that second house is other than Gemini and Virgo; in the twelfth, other than Taurus and Libra; in the seventh, other than Capricorn and Cancer. And in Aquarius or Leo, Mars produces no dosha in any house whatsoever. That is a detection rule rather than a cancellation, and the difference matters: it decides whether the dosha exists, not whether it is set aside. Which reading a report uses is one of the five differences catalogued in the five manglik answers.

When does one cancellation limit another?

More often than a flat list of rules can express, and the precedence is itself part of the tradition rather than a detail of implementation. This is where rule sets that look complete quietly go wrong.

Four separate rules fire on the condition “same rashi, different nakshatra”, which sounds entirely benign. But the OP, VSKP, SKU and RMP panchangas — set side by side in G.H. Visweswara’s comparison — name three specific nakshatra pairs that satisfy that condition and are nonetheless held to be forbidden: Bharani with Krittika in Aries, Pushya with Ashlesha in Cancer, and Dhanishta with Shatabhisha in Aquarius. A system that applies the four cancellations without the limit will cancel every dosha for precisely the three pairs the sources single out.

We know exactly what that looks like, because an earlier build of our own engine described the limitation in a comment and implemented nothing. A Bharani and Krittika couple scored a perfect thirty-six while being vetoed by the same rule set, in the same run. The registry has since been changed so the forbidden-pair rule names the four it suppresses and the evaluator reads that, which is a smaller fix than the failure deserved and a good argument for testing precedence rather than testing rules.

Rajju carries the same shape. Raman’s exception sets it aside when Graha Maitri, Rasi, Dina and Mahendra are all present — and the OP panchanga holds that a couple both in the Nabhi group is to be avoided regardless, which overrides the exception for that one group. Two sourced rules, pointing in opposite directions, with a stated order between them.

Which rules circulate with no source behind them?

The best-known one: that mangal dosha lapses once a person turns twenty-eight, sometimes thirty. It circulates widely and it appears in no classical text. Jagannath Hora states the position directly — no classical text specifies this age, and the belief most likely emerged from a misattribution of the Saturn return, which does occur at roughly that age and has nothing to do with Mars.

A rule of that kind is not a lenient reading of a tradition. It is a rule the tradition does not contain. The honest handling is to name it and say why it is excluded, rather than omit it silently and leave a reader wondering whether it was checked — so it sits in our registry as an explicitly excluded entry with the reason attached. That is a deliberate choice about which kind of mistake to risk: we would rather be visibly stricter on one popular rule than invisibly quiet about it.

What can a cancellation not do?

It cannot make the finding never have happened, and it cannot tell you anything about two people. A parihara operates inside one tradition’s arithmetic. It says: this school, holding this text, does not treat this configuration as blocking. That is a statement about a rule set. It is not a statement about a marriage, and no amount of citation converts it into one.

There is also a limit on how much leniency can buy you, and it is mildly deflating for anyone hoping a more lenient astrologer is the answer. Strictness about cancellations is the axis most people assume is doing all the work: the strict astrologer against the lenient one. It is real. Measured across the whole space of pairings, though, it is smaller than the axis nobody discusses — which system is being used at all. Taken together, the documented differences between five schools mean 56.4% of enumerated pairings trigger a tested difference or veto, and the strictness of the cancellation set is only part of that. A different reading may well produce a different verdict. The reason is more often a different rule table than a softer temperament.

And a cancellation that has been applied is still worth reading with the same scepticism as the dosha. A rule from a single popular source, applied to remove a finding from three classical ones, is a weaker document than the one it replaced — even though the answer it gives is the one the family wanted.

What should you ask for?

Two requests make any offered or refused cancellation examinable, and both are reasonable to put to any astrologer. Neither is a challenge, and neither requires you to know anything technical.

• The name of the rule and the text it comes from. A page in Muhurtha, a named panchanga, a verse. Not a summary of one. If the answer is that everyone knows it, that is itself an answer, and it puts the rule in the same category as the twenty-eight-year claim until a text turns up.
• Whether each rule was checked or could not be checked. If the birth time is uncertain, some of them cannot be, and knowing which ones is more useful than the total. It also tells you what a better birth time would actually buy, which is sometimes a great deal and sometimes nothing.

A cancellation with a citation behind it can be taken to a second astrologer and discussed on its terms. One without a citation can only be taken on trust, and a family that has been refused a marriage is rarely in a position to extend much more of that. If you want to see the conditions laid out with their sources before that conversation, the cancellation checker prints which rules fired, which did not, and which could not be evaluated on the data given.