Today’s Choghadiya — Eight Parts of Day, Eight of Night

Choghadiya means “four ghatikas”, and that is the whole of the arithmetic: the span from sunrise to sunset is cut into eight, the span from sunset to the next sunrise is cut into eight, and each of the sixteen pieces takes a name. Almanacs print the names as two tables of fifty-six cells and leave them unexplained, which is why they get copied rather than derived.

They are derived here, from one seven-element array. The seven choghadiya are the seven graha: Udveg is the Sun’s slot, Char is Venus’s, Labh is Mercury’s, Amrit is the Moon’s, Kaal is Saturn’s, Shubh is Jupiter’s, Rog is Mars’s. Read in that order the list is the Chaldean sequence — the graha ranked by apparent speed, rotated to begin at the Sun — and it is the same sequence that names the days of the week: step it by three and you get Sunday, Monday, Tuesday, Wednesday, Thursday, Friday, Saturday, because twenty-four hours is three places more than three full turns of a seven-name cycle.

The last column is what this site’s planet pages say each graha has charge of, and it is offered as a subject rather than as an outcome: Mercury’s slot is the one the tradition named for gain and Mercury is the graha it gives commerce to, and those two facts are the same fact told twice. The fourth column is the almanac’s three-way classification, carried in its own words — shubh is counted favourable, madhyam is counted middling, and specifically favoured for travel, ashubh is counted unfavourable. This page reports that classification. It does not add to it, and further down it shows a place where the same almanac disagrees with itself about it.

Given the seven, both cycles follow. The first slot of daylight belongs to the graha the weekday is named for — Sunday opens on Sun, so its first day slot is Udveg — and each slot after it advances one place along the cycle. Eight slots over seven names means exactly one name repeats, and it is always the opening one: the weekday’s own graha both opens and closes the daylight. The night runs the cycle backwards two places a slot and opens two places back from the day lord, which is measured off the almanac rather than argued from a principle. Those two rules reproduce all 112 cells of the printed tables, and the 112 cells are kept in the source as a fixture so the build fails if the rules and the almanac ever part company.

The same cycle at a different gear ratio gives the planetary hora: twelve pieces of daylight instead of eight, one step of the cycle each. Because twelve horas take you five places along a seven-name cycle, and five places forward is two places back, the first slot of the night and the thirteenth hora share a graha on all seven weekdays — just as the first slot of the day and the first hora do. Choghadiya and hora are not two systems. They are one system counted in eighths and in twelfths.

Choghadiya guides say a period runs about ninety-six minutes. Ninety-six minutes is a twelve-hour day divided by eight, which happens on the equinox and on no other day of the year. A slot is a fraction of the actual daylight, so it is as variable as the daylight is. Over the 729 sunrise-to-sunrise spans this site’s ephemeris table brackets for New Delhi — 1 January 2026 to 30 December 2027 — here is the range:

Those figures are exact rather than rounded. Sunrise and sunset are tabulated to the whole minute, so a day length is a whole number of minutes and an eighth of it always terminates within three decimal places. The widest a day slot ever gets is 104.75 minutes and the narrowest is 77.375 — a spread of 27.375 minutes, which is 26.1% of the longest one. Ninety-six minutes overstates the day slot on 453 of the 729 days and understates it on 276.

The day half is the longer of the two on 391 of 729 days, and the lead changes hands 4 times inside the table: 16 March 2026, when the day half takes the lead; 28 September 2026, when the night half takes the lead; 17 March 2027, when the day half takes the lead; 28 September 2027, when the night half takes the lead. Those dates sit a few days off the equinoxes, and the offset is not an error. Sunrise and sunset are defined here at a solar zenith of 90.833 degrees rather than 90 — the standard allowance for refraction and the Sun’s own width — so the visible day runs a few minutes longer than the geometric one at both ends. Daylight therefore passes twelve hours a few days before the March equinox and does not fall back under it until a few days after the September one, which is where these four dates land. The closest the two ever come is 27 September 2026: 90.125 minutes of day against 90 of night. The widest is 14 June 2026, at 104.75 against 75.25.

Sixteen slots do not tile a flat 1,440 minutes. Across the whole table their total takes 4 values — 1438, 1439, 1440, 1441 minutes — because each of the three times involved is truncated to the whole minute and the day is not exactly twenty-four hours long anyway. The spread is the rounding, not the sky.

Two boundaries need no ephemeris at all. Four eighths is half the daylight and six twelfths is half the daylight, so local apparent noon — the midpoint of sunrise and sunset, which is not 12:00 — is simultaneously the seam between the 4th and 5th choghadiya and the seam between the 6th and 7th hora. That holds on all 729 days of the table, at any latitude, and is checked rather than asserted. And Abhijit, the eighth of fifteen parts of daylight, is 8/15 of a choghadiya slot exactly — 53.3% of it — sitting astride that same midpoint, so it always overlaps the two slots the seam divides.

It does not tell you when to act. The three class words are the almanac’s and are printed as the almanac’s; the section above shows the same almanac putting a favourably-counted eighth inside a window it elsewhere asks you to leave alone, on eight of twenty-one weekday-and-window pairs. A page cannot show you that and then also hand you a rule.

It does not compute a minute. Sunrise and sunset come from a generated table built on the JPL DE421 ephemeris; every boundary on the live chart is arithmetic on those two numbers and nothing else. When the table does not reach the date, the chart says so and leaves the times blank instead of assuming a six o’clock sunrise, which is what this page used to do all year round.

And the clock times are New Delhi’s. Longitude moves every boundary: sunrise in Kolkata is nearly an hour earlier than in Ahmedabad, so a slot that opens at 9:12 in one opens at 10:05 in the other. The names, their order and every fraction on this page are the same everywhere; only the minutes are local. The panchang pages carry sunrise for New Delhi and 37 other cities, with Rahu Kaal, Gulika Kaal and Yamaghanta as the eighths this page has just identified.