Bhadra is one thing and one thing only: the Vishti karana running. Everything else said about it is interpretation laid on top of that.
A karana is half a tithi: six degrees of the Moon's separation from the Sun. Vishti is one of seven karanas that cycle. So Bhadra arrives on a schedule you can compute from two positions, and it arrives far more often than its reputation suggests.
Eight times a month, not twice
Here is the arithmetic, because it is short and it settles the question.
A lunar month contains sixty karanas. Four of them are fixed (Shakuni, Chatushpada, Naga and Kimstughna) and appear once each, around the new moon. The remaining fifty-six are the seven movable karanas repeating eight times over: Bava, Balava, Kaulava, Taitila, Gara, Vanija, Vishti.
Vishti is the seventh of the seven. So it occupies eight of every sixty karanas, which is 13.3% of all time. About ninety-nine windows a year, each around twelve hours long.
That figure is worth sitting with, because Bhadra is usually described as something to watch out for. It covers an eighth of the calendar. Any rule that avoids it is ruling out roughly one day in eight by construction, and a collision with it is not remarkable.
I got this wrong myself while building the tool. The first version of the search assumed roughly two windows a month and returned a plausible-looking list that was a third of the real one. It was caught by checking the total against the 13.3% the karana arithmetic predicts, not by looking at the dates, which would have seemed perfectly reasonable.
Where it does real work
The frequency is not an argument that Bhadra never matters. There is one observance where it demonstrably moves a date rather than merely delaying an hour.
Holika Dahan must avoid Bhadra, and Bhadra runs from the start of Purnima in every year I have checked, so it routinely swallows the entire pradosh window in which the bonfire should be lit.
Usually that only delays the lighting. In 2024 Bhadra cleared at 23:13 and in 2025 at 23:27, both still that night, so the dahan happened late on the same date.
In 2026 it runs until 05:28 the following morning. There is no usable evening left, and the observance moves to the next day. The festival calendar applies that test.
Raksha Bandhan carries the same constraint on the thread ceremony, which is why the usable window in some years is confined to a couple of hours, or to the morning only.
What this page does not compute
Two things, both deliberately.
Bhadra vasa. The doctrine that Bhadra "resides" in one of three lokas depending on the tithi and paksha, and only troubles the earth when it resides here. The rule circulates in several incompatible forms (keyed sometimes to the rashi the Moon occupies, sometimes to the paksha and tithi) and the versions do not agree on which case is which. It is absent rather than half-applied.
Bhadra mukha and puchha. The division of a Bhadra window into a "face" to be avoided and a "tail" said to be auspicious. The proportions differ between sources, and applying one set silently would put a specific time on the page that another source would put elsewhere.
What is computed is the karana, which is unambiguous: half a tithi, six degrees of elongation, the same instant everywhere on earth.
Why the city barely matters here, and the ayanamsa not at all
A karana boundary is a single instant worldwide. Bhadra begins and ends at the same moment in Varanasi and in Vancouver; only the clock time on the wall differs. So the city selector on this page changes the display rather than the event, which is unusual for this site.
The ayanamsa does not move it either. A karana is defined by the difference between two longitudes, and an ayanamsa shifts both by the same amount, so the boundaries come out almost identical under Lahiri, Raman and Krishnamurti.
Where sources genuinely disagree is on the vasa and the mukha/puchha divisions above, not on when Vishti runs.
What the seven movable karanas are, and why Vishti is one of them
The karana is the least discussed of the five limbs of the panchanga, and its structure is simple enough to state completely.
Each tithi is divided in half, giving sixty karanas to a lunar month. Eleven karanas exist by name. Seven of them (Bava, Balava, Kaulava, Taitila, Gara, Vanija and Vishti) rotate continuously through fifty-six of the sixty positions, which is exactly eight full cycles. The other four (Shakuni, Chatushpada, Naga and Kimstughna) are fixed: they occupy the last three positions of the month and the first one, clustered around the new moon, and they appear once each.
So there is nothing irregular about when Vishti arrives. It is the seventh seat on a wheel that turns once every seven half-tithis, and it comes round eight times a month for exactly that reason.
That is also why a Bhadra window is around twelve hours: it is one karana, and a karana is half a tithi. The variation from about ten to about thirteen hours is the same variation that makes tithis run from twenty to twenty-six hours, which is the Moon's changing speed through its orbit.
Checking the page against itself
The frequency is the check worth doing, and it is the one that caught my own error.
Count the Bhadra windows in a year and you should get somewhere near ninety-nine. Add up their durations and you should get near 13.3% of the year. If either figure comes out low, the search has truncated; if either comes out high, something is being double-counted.
Neither number can be verified by looking at a list of dates, which all look equally plausible. They can only be verified against the karana arithmetic, and that is the argument for stating the arithmetic on the page rather than only the times.
Why the frequency is worth publishing at all
There is a general point here that goes beyond Bhadra, and it is the reason this site prints base rates next to its results.
A period that covers 13% of the calendar will coincide with roughly 13% of anything: good days, bad days, journeys that went well, journeys that went badly. Noticing that something unfortunate happened during Bhadra is therefore not evidence of much on its own, because the same is true of one day in eight by construction.
That is not an argument that the observance is pointless. It is an argument that the frequency belongs on the page alongside the times, so that a reader can tell the difference between a rare alignment and a regular one. Most listings give the times and omit the frequency, which leaves an impression of rarity that the arithmetic does not support.
The same reasoning appears on the Panchak page, where the figure is a sixth of all time, and on the Kaal Sarp page, where the base rate is the whole point.
Reading the window list
The page gives every Bhadra window over the following thirty days, with start and end instants and the duration.
Those instants are the karana boundaries, computed from the Sun and Moon rather than looked up, and they are the same worldwide. What changes with your city is only the clock time they are displayed in.
The current status is read at that morning's sunrise, in keeping with how every other element on this site is reported for a day. If Bhadra begins at eleven in the morning, the day is reported as not being in Bhadra at sunrise, and the window is listed separately with its own times. That is more useful than a single yes-or-no for a period that routinely starts and ends mid-day.