Why the “9 hours” and “12 hours” before an eclipse don’t match the eclipse times

1 What the rule actually says

The pre-eclipse period is Vedha (વેધ). During it, food is not taken — Vedha and the eclipse food-stop are the same boundary. Its length is given in prahars (also called yam), not in hours:

And a prahar is defined by sunrise and sunset at your own location:

2 So where do 9 and 12 come from?

From the nominal prahar. A full day-and-night is 24 hours, divided into 8 watches, so a prahar is loosely quoted as 3 hours. On that reckoning:

That arithmetic is only exact when day and night are both exactly 12 hours long — i.e. at the equinox. Every other day of the year, a prahar is not 3 hours, and the saying drifts from the actual rule. The further from the equator you are, the more it drifts.

3 Reason one — prahars stretch and shrink

Two real days at the same place (Bolton, UK), as computed by the engine:

A solar eclipse asks for four prahars. In Bolton in summer, four day prahars come to 15h 00m. In winter they come to 7h 32m. Neither is 12.

4 Reason two — it starts at a boundary, not a countdown

This is the part that surprises most people, and it is the single biggest reason the numbers don’t line up. You do not subtract 4 prahars from the eclipse time. The procedure is:

Because you land on the start of a watch, the gap always also includes however much of the contact prahar had already run before the eclipse began. That is an extra 0 to 1 whole prahar on top of the nominal 3 or 4 — and it varies with every eclipse.

5 Reason three — first contact, not maximum

Published tables usually headline the maximum of the eclipse, because that is the photogenic moment. The rule anchors on first contact (sparsha) — the instant the shadow first touches the disc. For today’s eclipse that is a difference of 57 minutes. Measuring back from the wrong anchor shifts everything by that much again.

6 Worked example — the solar eclipse of 12 August 2026

Computed for Bolton, UK. Sunrise 05:44, sunset 20:45 (BST) — a long summer day, so day prahars are long (3h 45m) and night prahars are short (2h 15m).

First contact is at 18:13, which falls in Day Prahar 4. Counting back four prahars — Day P3, Day P2, Day P1, then the previous Night P4 — the fast begins at the start of that night watch: 03:29.

Where the 14h 44m comes from

Notice the third row. Even if every prahar had been exactly 3 hours, you would still not get 12, because the fast begins at the start of Day Prahar 4 — well over an hour before the eclipse touched anything.

7 Reason four — it is your sunrise, not India’s

Prahars are built from local sunrise and sunset, so the same eclipse gives a different food-stop in different cities. And there is a prior question: if the eclipse is not visible where you are, Vedha and Sutak do not apply at all (no Rahu-darshan). The engine checks visibility first, before it computes any timing.

8 How far off is the saying, in practice?

Every visible eclipse for Bolton, UK over the next six years, straight from the engine:

Two rows are worth a second look.

2 August 2027 lands on 11h 59m 54s — six seconds under the “12 hours”. That is a coincidence, not a confirmation. The day prahar that morning happens to be 3h 55m and first contact happens to fall early in Day Prahar 1, and the two errors cancel. Change the date or the town and the agreement disappears.

1 June 2030 is the opposite. The solar Vedha is 8h 03m — shorter than the lunar “9 hours”, even though solar is the longer rule at 4 prahars. First contact is just after sunrise, so the count runs back through four short midsummer night watches of 1h 49m each. The shorthand cannot express that; the prahar rule handles it without any special case.

9 The concession for those who cannot fast

For children, the elderly and the unwell, the lookback is one prahar rather than the full three or four. Everything else about the procedure is unchanged — and that includes section 4: you do not subtract one prahar from first contact. You find the prahar containing first contact, step back one watch, and begin at the start of that watch. The food-stop lands on a boundary here exactly as it does for the full Vedha; no food-stop ever falls mid-watch.

For the 12 August 2026 eclipse in Bolton, first contact at 18:13 sits in Day Prahar 4. One watch back is Day Prahar 3, which begins at 13:14 — that is the concession time, against 03:29 for everyone else. Note it is 4h 59m before contact, not 3h 45m: the same “extra part-prahar” from section 4 applies, so even the concession is longer than one bare prahar.

And when they may eat again

The concession is symmetric — it shortens the fast at both ends. Ordinarily everyone eats again at the eclipse release (moksha), after snan. But when the body goes down still eclipsed — ગ્રસ્તાસ્ત, grastasta — the ordinary observer must wait for it to come back up clear: the next moonrise for a lunar eclipse, the next sunrise for a solar one. That can be a further half-day of fasting.

A child, an elderly person or someone unwell does not wait. Their restriction ends when the eclipsed body sets, because at that moment the darshan is over. The same visibility principle that governs section 7 governs the release: no Rahu in the sky, nothing to observe.

10 What this means in practice