Sunrise is a decision, not an observation, the gist

Sunrise is a decision, not an observation

There is no single moment called sunrise, only four different moments that people have good reasons to call sunrise, and at Ujjain on one August morning they are spread across three minutes and forty-two seconds.

This is the short way in. The essay has the working, the sources and the sky.

By one of the definitions in everyday use, sunrise happens when the sun is already fully up. The whole disc is above the horizon, with its lower edge clear of the ground by half its own width, and only then does the almanac agree to call it sunrise.

Ask a computer when the sun rises and it cannot answer until you tell it two things.

The first is which bit of the sun you mean. The sun is not a dot. It is a disc about half a degree wide, so its top edge and its middle are about sixteen arcminutes apart. An arcminute is a sixtieth of a degree, a small angle: the whole sun is about thirty-two of them across.

The second is whether the air counts. Air bends light. Light from a low sun comes in at a slant through thick air, gets bent downwards, and arrives from a direction slightly higher than where the sun really is. So the sun looks lifted. At the horizon the standard allowance for that lift is thirty-four arcminutes, which is more than the width of the sun itself. When you watch the sun sitting on the horizon, the real sun is already below it.

Two questions, two answers each, four sunrises.

The essay works all four for one real place and morning: Ujjain, 28 August 2026. The times are 06:07:39, 06:08:50, 06:10:10 and 06:11:21. Getting from arcminutes to minutes of clock takes one number, worked out in the essay: at that place on that date the sun climbs 13.5 arcminutes a minute, so each arcminute of height is worth 4.43 seconds.

The sixteen is steady. The thirty-four is not. It is a figure for average air, and real air is warmer, thinner, colder or stiller than average, which shifts the answer by seconds and sometimes by a minute.

Then the essay asks what the three minutes actually cost, because in this calendar a day begins at sunrise, and which day a fast falls on is settled by what is happening at that minute.

Words decoded

sunrise, as computed: the moment a stated part of the sun reaches a stated height above a stated horizon; not a sighting
arcminute: a sixtieth of a degree; the sun's disc is about 32 arcminutes wide
semi-diameter: the distance from the middle of the sun to its edge, about 16 arcminutes
upper limb: the top edge of the sun's disc
refraction: the bending of light by air, which lifts everything near the horizon above its true position
true (or geometric) altitude: the height something really is above the horizon, with the bending taken out
apparent altitude: the height it looks, with the bending left in
declination: how far north or south of the equator the sun stands on a given day
hour angle: how far the sun stands from the north-south line overhead
dip of the horizon: how far the visible horizon sits below eye level when you are high up
tithi (तिथि): the lunar day the Indian almanac counts by
panchang (पञ्चाङ्ग): the Indian almanac, printed or online
sūryodaya (सूर्योदय): the Sanskrit word for sunrise, the sun's rising

Why anyone should careFast days and festival dates are decided by which lunar day is running at sunrise, so the sunrise minute is doing real work in a lot of households. Two almanacs can print sunrise times three minutes apart without either being wrong, and roughly once a year at a given city those three minutes are enough to put the same fast on two different dates. Knowing that stops the argument being about who is right. It becomes a question anybody can ask and anybody can answer: which sunrise is this sheet using.

Read it properly, 8 min Sunrise is a decision, not an observation