The reference-point problem nobody explains

The part of the manglik rule that everybody argues about turns out to be the part that matters least. There is a dispute anyone can find in an afternoon: whether the 2nd house counts, whether the 1st does, whether the 5th belongs. Forums argue it, software vendors quietly pick a side, and families inherit whichever answer their region settled on.

Inside the same calculation there is a second choice that almost nobody argues about, largely because most people have never been told it is a choice. Separated and measured one at a time, the house list moves the flag rate by about 3 percentage points and the second choice moves it by about 29.

Which half of the rule does the internet argue about?

The house lists, which really do differ, and which are the narrower of the two levers by a wide margin. The disagreement is properly sourced and it is not going to be resolved: one enumeration includes the 1st house and excludes the 2nd, another does the reverse, merged sets hold both, and a single set adds the 5th.

Strip out the houses every list shares and what remains in dispute is three houses out of twelve. That is a genuine disagreement, and for a chart with Mars sitting in one of those three it decides everything. Across a corpus of charts, though, it is a narrow lever, because Mars has to land in one of three specific houses before the choice of list changes anything at all.

Where those enumerations came from, and why no surviving text reconciles them, is the business of the piece on the five readings. This page takes the lists as given and asks a different question: how much of the total spread do they actually account for?

What has to be true before any house list can be applied?

Something has to be counted from, because a house number is a distance and a distance needs an origin. That requirement is so basic it disappears from view, which is precisely why it is never stated on a report.

One convention takes the origin as the ascendant. Houses are counted from the rising sign, Mars lands somewhere in that count, and the list is consulted once. Clean or flagged, and the procedure is over.

Another convention counts from the ascendant, then counts again from the Moon, then again from Venus. Three origins, three different house numbers for the same planet, three consultations of the same list. If any one of them lands on a counted house, the finding is raised.

Both conventions are traditional and both are cited. The second is not a stricter reading of the first — that is the thing worth being precise about. It is a structurally different procedure, because it gives Mars three chances to land rather than one, and a procedure with three chances is not the same procedure run more severely.

Where does the three-origin instruction come from?

From a separate verse that names the three points in a single compound, transmitted alongside the house enumerations rather than inside them. The southern lineage carries a line that opens lagnendu shukrat — from the lagna, the Moon and Venus — and it is that line, not the house verse, which establishes that the test runs three times.

The distinction matters more than it first appears. The enumeration verses transmitted in the Parasara material and in the Manasagari say which houses. A different verse says from where. They are separable instructions, and because they are separable they can be, and have been, recombined.

Which is how the awkward middle cases arise. Nothing in any source tells a reader to take one lineage’s house enumeration and apply the other lineage’s origins to it. But the two instructions arrive as separate lines, so pairing them is a two-second decision in a configuration file, and the result looks exactly as sourced on screen as a pairing with centuries behind it. Both halves are cited. The combination is not.

How were the two axes measured separately?

By holding one still and moving the other, which is the only way to find out what an axis is worth on its own. First the origin convention was fixed and only the house list was swapped. Then the house list was fixed and only the origins were swapped. Each run reports what that axis alone contributes.

Swapping house lists moves the manglik rate by roughly 3 percentage points. Swapping the origin convention moves it by roughly 29.

The axis carrying the entire public argument is worth about a tenth of the axis nobody mentions.

The corpus is the same one throughout: we ran every documented set over one fixed collection of 2000 charts, changing nothing between runs but the setting under test. That part of the census is a sample rather than a population, because a manglik test needs an actual position for Mars and therefore a date, a time and a place. The star-based half of the same census is exhaustive, and we say which is which rather than letting one borrow the other’s authority.

That attribution is what makes the headline spread possible in the first place. The flag rate on identical charts runs from 11.7% under Astro-Vision (serious Kuja dosha only) to 73% under OP / SKU panchangas, and most of that distance is not house lists. It is how many points in the chart Mars is judged from.

Why does adding an origin move the rate so far?

Because the passes combine as an OR rather than an AND, and an OR only ever grows. Under a single-origin convention Mars is either in a counted house or it is not: one test, one outcome. Under a three-origin convention the chart is clean only if Mars misses the counted houses from the ascendant, and misses them from the Moon, and misses them from Venus.

Suppose a single pass flags roughly a third of charts. If the three passes were independent, the chance of missing all three would be about thirty per cent, so nearly seven charts in ten would be flagged — from the same house list, with no doctrinal change whatever. Three passes turn a minority finding into a majority one by arithmetic alone.

The three passes are not fully independent, which is why the real figures are not quite that clean. The Moon and Venus and the ascendant are not scattered at random with respect to one another; Venus stays close to the Sun, and both the Moon and the ascendant sweep the whole zodiac but at wildly different speeds. The correlations pull the combined rate a little away from the independent estimate. They do not change its direction, because nothing can: adding an origin can only push the rate up.

Set that against the house-list lever. Adding one house to a list of five widens the target by a fifth. Adding one origin aims at the same target a second time. Those are not comparable operations, and the measured gap between the two axes is what you would expect once you notice they are not.

Do the traditions themselves rank the origins?

They do, and two widely cited sources rank them in opposite directions. This is the part of the subject that gets least attention and deserves considerably more, because it means the second axis is contested inside the tradition and not only between traditions.

Raman, in Muhurtha, describes the finding as weak when read from the ascendant, a little stronger from the Moon, and more powerful still from Venus. Read plainly, that puts the ascendant at the bottom of the ranking and Venus at the top.

Several panchanga traditions state the reverse. In their formulation the finding from the Moon and from Venus carries the same weight as each other, and the finding from the ascendant carries double. Same three origins, same word for the finding, an ordering turned upside down.

Neither source publishes a coefficient. Both state an ordering in words, and any engine that turns those words into weights is choosing numbers the source never gave it. We do carry both orderings, and we flag them as ordinal rather than arithmetic for exactly that reason — a weighted score presented as though the weighting were sourced is a small fabrication that looks like precision.

An opinion, offered as one: this contradiction is more damaging to the idea of a single correct manglik verdict than the house-list dispute has ever been. Two respected sources cannot even agree on which reading of the same chart is the serious one.

There is a practical consequence for anyone holding a report. A verdict raised only from Venus is, on Raman’s ordering, the strongest form of the finding and, on the panchanga ordering, the weakest. Neither reading is a misprint and neither is fringe. So a family told that the finding is serious, or told that it is mild, has been given a judgement that depends on which of two orderings the reader was trained in — and that dependency is invisible on the page, because severity language rarely names its origin at all.

Is there a reading that drops the ascendant entirely?

Yes, and it is the clearest demonstration that a table of house numbers is not enough to identify a tradition. One documented set counts from the Moon, from Venus, and from the lord of the 7th house. There is no ascendant in it at all.

Its house list is identical, number for number, to a common merged set. Print the two side by side as rows of houses and they are indistinguishable. Run them on the same chart and they can disagree, because one of them is looking at the chart from a point the other never uses and ignoring a point the other starts from.

That is why the figure on these pages carries the origins next to each row rather than the houses alone. A chart of house lists shows the argument everybody is already having and hides the one that decides the answer, which makes it worse than no chart at all: a table looks more authoritative than a sentence, so a misleading table is believed harder.

Why is the lord of the 7th a different kind of origin?

Because it is derived rather than fixed — it is not a point in the sky but a planet chosen by the chart itself. The ascendant, the Moon and Venus are all positions you can look up. The lord of the 7th is whichever planet rules the sign occupying the seventh house, so identifying it requires the ascendant first, and then a lordship table.

The consequence is that this origin changes identity from chart to chart. In one chart it might be Venus, in which case the reading collapses into a two-origin test. In another it might be Mars itself, which raises a question the sources do not obviously answer: what does it mean to count Mars in houses measured from Mars? A rule written for a room with an astrologer in it can leave that unresolved. Software cannot.

It also inherits the birth-time uncertainty of the ascendant, indirectly. Change the rising sign and the seventh sign changes with it, and so does its lord, and so does the origin. A reading that looks independent of the birth time is quietly not, and the way to find out is to check what the ascendant actually was before trusting anything counted from it.

Does the origin decide the answer for every chart?

No, and the attribution above is a rate across many charts rather than a claim about any one of them. For a particular chart the house list may be the only thing that matters, or the origin may be, or neither may.

The clean case is worth stating because it is common and it is reassuring in a way most of this subject is not. If Mars falls in the 4th, 7th, 8th or 12th house counted from the ascendant, every documented list counts it and no additional origin can reverse that. The verdict is stable, a second reading will agree, and there is nothing to be gained by asking a third person.

The origin decides the outcome for charts that come back clean on the first pass. Those are the charts where a second opinion can genuinely change the answer, and they are also the charts where a family is most likely to be told two different things by two people who are both being careful.

Underneath all of it sits the ordinary business of getting the chart right at all: the recorded time resolved against the offset that clock was running on, using something like the IANA time zone database, and positions traceable to a published ephemeris and checkable against JPL Horizons. None of that is where the disagreement lives, but a chart wrong in the inputs makes every argument above irrelevant.

Which line on a report reveals the origin?

A phrase like “from lagna”, “from Chandra” or “from Shukra” beside the verdict, or a table with one row per reference point. A detailed report usually prints a house number. It much less often prints the origin that number was counted from, and it very rarely prints every origin that was tested.

If the report has that table, the reasoning is fully visible and can be checked against any house list you like, including one from a different tradition. If it has only a verdict, the single most useful question to send back is whether the test was run from the Moon and from Venus as well as from the ascendant. That answer is one sentence long, and on the measurement above it is worth about ten times as much as the argument the internet is having.

Reading a report for what it does and does not disclose is a skill worth having generally, and we set out the whole checklist in the guide to reading any matching report. For this axis in particular, the tools print every origin they tested rather than the one that produced the headline: the manglik checker shows one row per reference point per tradition, and the kuja dosha view does the same under the southern vocabulary.

What does this attribution not prove?

It does not identify a correct convention, and it was not built to. Three limits are worth stating plainly rather than leaving for someone else to find.

• The traditions are not being scored against one another. A lineage counting three origins is not overcounting; it is applying its own rule faithfully. The measurement describes how far apart the rules land, not which rule is sound.
• The corpus uses dates spread across several decades and a handful of Indian cities. A real birth cohort has a different age and geography profile, so these are the shape of the spread rather than a demographic statistic about anybody.
• Cancellation rules sit on top of everything here, and those registries differ between traditions too. The rates discussed on this page are measured before that layer is applied, which makes them the widest version of each rule rather than its final one.

What survives all three is the finding this page exists for, and it is a small one stated carefully: of the two settings inside the manglik calculation, the one that is universally argued about is the one that matters less, and the one that decides most of the outcome is almost never written down. How far that reaches beyond manglik is a separate question with its own measurement, and the pattern is not confined to Mars.