How we test our prayer times

What we found

• The sun is where we put it. On 2 November 2026 we checked sunrise, noon and sunset at 18 cities for every day of 2027 against NASA JPL’s Horizons. The largest gap was 2.7 seconds.
• Our pages show what our code calculates. On 3 November 2026 all 690 month-table cells and 23 “today” blocks we checked matched it exactly, with the device set to two different time zones.
• Other prayer-time sites differ for stated reasons. On 3 November 2026 we compared 1,350 times at 25 cities on 3 dates with two prayer-time websites and one calculation service. 836 were within 2 minutes. Each larger gap was traced to a cause we could reproduce: a different method or Asr rule, minutes our UK setting adds, a mosque’s own table, or height above sea level. With our method set, the calculation service matched 162 of 162 Australian times within 2 minutes.
• The astronomy is close. Over 9,125 city-days, adhan-js and a separate NOAA-formula check differ by at most 1.62 min at sunrise, 1.27 at sunset and 0.71 at noon.
• Sunrise and Maghrib aren’t where timetables differ most. Fajr and Isha are. Against mosque timetables read across dates, the UK starting setting averages 3.1 minutes at Fajr from London Central Mosque (13 dates, hand-read; at most 6) and 9.5 from Birmingham Central Mosque (365 days; at most 22). One-day reads of local mosque pages were further off: Leeds 16, Liverpool 15, Nottingham 21, Glasgow 24 minutes.
• UK timetables disagree with each other by much more than that. Wifaqul Ulama’s PDFs sit an average of 42 to 73 minutes from our UK starting setting at Fajr, because they use a different summer limit. A setting built to mirror them comes within about 1 minute.
• Presets don’t always match the body they’re named after. See the table below. We don’t call any setting by a body’s name on the page, unless we say which test it passed and where.

Why these starting settings

UK cities below 55° north start on “Seasonal minutes from sunrise and sunset (18° only if nearer)”, a stated starting point and not a choice between sources. Fajr is set by the seasonal minutes, not by 18°, on most or all days. The UK evidence for it is the mosque timetables named above, and each city page lists only that city’s own. It is often the later Fajr: on 194 to 217 days of 2027, depending on the city, it is later than Diyanet, AlAdhan and (where read) Wifaqul Ulama alike.

Glasgow and Edinburgh have no starting setting. Above 55° north the same formula changes and jumps, and we found nothing local that supports one setting there, so those pages ask you to choose. Read why.

Australian cities start on 18° / 17° with the twilight-angle limit, except Sydney. Sydney starts on Fajr 18° with Isha 90 minutes after sunset: two Sydney sources we read sit about 8 minutes from 18° / 17° at Isha on average, and under a minute from this. Our one multi-date body check for 18° / 17° is Melbourne, against the Islamic Council of Victoria, within 1 minute through the year. Other Australian bodies differ.

Rounding

Times are rounded to the nearest minute, like the library’s output. Unless the method adds minutes, Maghrib can show up to 30 seconds before the calculated sunset, and Sunrise up to 30 seconds after it. The UK starting setting adds 5 minutes to Dhuhr and 3 minutes to Maghrib. We haven’t changed this rounding.

No scholar or imam has reviewed this site, and none chose these settings. You can change them on every page that shows a place’s timetable, apart from Match my mosque, which finds the nearest setting for you. Lists of places (All places, Other cities, the Near me list) always show each place’s starting setting.

How we report closeness

• Each city names the timetables we compared and how many days each covered.
• Where a city has two or more, we name the furthest as well as the closest, so a close match isn’t shown on its own.
• A one-day read is a warning, not a score.
• Times are for sea level. Each city’s height would move sunrise and sunset by roughly 0.5 to 4.3 minutes in our cities, our estimate from that height; each city page gives its own figure.

What we couldn’t test

• East London Mosque’s own site refused our automated browser. London’s Unified times were read from London Central Mosque’s calendar (12 dates, by eye) and one day from londonprayertimes.com.
• Only one day for the mosque pages we read in London, Glasgow, Birmingham, Sheffield, Leeds, Edinburgh, Nottingham, Sydney, Melbourne and Brisbane. One day is a warning, not a score.
• No local mosque page read for 14 of our 25 places: Bradford, Manchester, Luton, Leicester, Coventry, Oldham, Cardiff, Blackburn, Perth, Adelaide, Canberra, Gold Coast, Darwin and Hobart. These are marked as gaps on their pages. Of those, Perth, Adelaide, Canberra, Gold Coast, Darwin and Hobart have no local source at all, so they rest on calculators and, for the first 25 places, Diyanet.
• Sources’ own years differ: Diyanet, the Moonsighting calculator and AlAdhan are 2027; UK PDFs and most mosque data are 2026. Compared date by date, for the year each publishes.
• AlAdhan and the Moonsighting calculator are software, not authorities: they show that independent code agrees, not that a timetable is followed.
• US Naval Observatory data was unreachable from our test machine, so the astronomy check uses NOAA’s published formula instead.
• Places outside our list have pages too, hidden from search engines and calculated the same way. None has been compared with a local timetable, and each of those pages says so.
• Real iPhone and Android calendars and screen readers on phones. Headless Chromium only.
• Whether a close match is independent. Reading Mosque’s and Harrow Central Mosque’s timetables sit within 0.1 minutes of our seasonal setting. That can mean they are made with the same formula, not that two sources agree on their own.
• Whether a mosque prays at its printed start times. We compare start times only, not jamaat times.

Corrections log

• 4 November 2026, before launch, closest-timetable figures: every page now reads one generated set of figures. Three moved by 0.1 minutes: Luton’s Wifaqul Ulama Fajr 43.4 to 43.5, Oldham’s 48.4 to 48.5, and Melbourne’s Islamic Council of Victoria Isha 0.2 to 0.1. No time changed.
• 31 October 2026, before launch, Glasgow and Edinburgh: these pages started on a setting whose Fajr was the latest offered, with a jump at 55° north. Now there is no starting setting. Our own separate check found it. That check isn’t an outside or religious review.
• 31 October 2026, before launch, UK label: “18° / 18°” was the label, but seasonal minutes set the time. The label now says so.
• 1 November 2026, before launch, label from 55° north: the seasonal method was called “one seventh of the night”, but from about October to March its seasonal minutes set the time. The label now names both. No time changed.
• 1 November 2026, before launch, wrong Isha on some summer nights: near the sun’s limit for an angle, the calculation library could give a wrong Isha, and rarely a wrong Fajr, in some cases hours off and still in the usual order. Our earlier check of order couldn’t see most of these. No starting setting at our listed cities was affected from 2026 to 2030; optional settings were, most under “Middle of the night”. On those days we now show no time for that prayer and point to your mosque’s timetable. Every other time stayed the same.

Each later correction will be listed here with its date, cities, cause and fix.