Time is required for three main classes of use : scientific, requiring a uniform scale of exact SI seconds; astronomical, requiring knowledge of the movement of the Earth; and civil, requiring the usual "multiples" and subdivisions of the solar day, and perhaps Summer Time. Astronomy also uses the sidereal day.
For ordinary computing purposes, we only need to consider the effects of location ("Greenwich" or other) and season (Winter or Summer Time); and possibly of Leap Seconds.
I do not deal with the measurement of time or of the Earth's orbit, rotation, etc., nor with details of scales such as TT & UT1; only with time as ordinarily used. For metrologist-grade definitions of time scales, seek within Date and Time Index and Links; try the NPL, RGO/ROG/NMM, USNO, BIPM, and IERS Websites.
Some links to Web current-time displays are in Astronomy / Astronautics 1.
For Measurements and Units, see in Kaye & Laby Online.
Originally, local solar time, for the location or for the nearby town or city, was used.
During the course of a year, the middle of a solar day does not fall at equal intervals of time (the Equation of Time). A uniform civil time scale is better suited to mechanical clocks; so a fictitious average uniformly-moving sun, the Mean Sun, is used instead of the actual sun. Hence the term "Greenwich Mean Time", known as GMT.
In the UK, GMT was first adopted by the GWR in 1840, by our railways as a whole (and the Post Office) in 1847, by the UK in general in the eighteen-fifties, and by UK Law in the Statutes (Definition of Time) Act, 1880 (43 & 44 Vict. c. 9), Royal Assent 1880-08-02. See Myers, Howse, etc.
Since 1880, legal time in Great Britain has been GMT-based (in Ireland from 1916), though time signals actually give UTC. English forms of EU rules use GMT.
The TAI second is the SI second; TAI has no leap seconds. The UTC second is the SI second; UTC is adjusted by Leap Seconds to agree with solar mean time. GMT has 24×60×60 = 86400 seconds per average solar day and per calendar day. GMT and UTC never differ by more than 0.9 seconds.
UTC is not a true acronym, but is a compromise between ""Universal Cordinated Time" and "Temps Universel Coordonnée".
Universal Time (UT) is a general term for a date/time, in any era, which matches present GMT.
GMT/UTC are sometimes denoted by Zulu (phonetic alphabet : Z).
The start of the day has not always been at midnight.
Until 1925, astronomers used the term GMT for a Greenwich Mean Astronomical Time, GMAT, starting at noon of the same date by civil GMT.
For more, see references in Date and Time Index and Links.
North American railroads introduced (zoned) standard time there on 1883-11-18 Sun; most larger US towns were using it by October 1884; legalized by US Act of Congress in 1918.
New Zealand may have been the first to introduce a national Standard Time, on 1868-11-02, of GMT+11:30.
Since this site is for general use, I tend to use UTC to mean the present UTC (with leap seconds), and GMT for time which has exactly 60×60×24 seconds every day and agrees with London Winter Time. I understand that UT (Universal Time) is, for experts, a better term for the latter; but it seems to me that it is too likely not to be understood, or to be taken as a typo.
The SI Second is now the fundamental unit for the measurement of duration. It is defined* to considerable precision in the International System of Units (SI), by reference to an atomic resonance frequency; see also ISO 31-1 and The International System of Units (SI), PDF from BIPM.
The proper SI abbreviation for the second is "s" - hence the symbols fs, ps, ns, µs (or us), ms, s, ks, Ms, Gs, Ts, etc. It is not "S", which is for the Siemens, the unit of electrical conductance. That unit is named after Werner Siemens, so a single unit is One Siemens.
* : La seconde est la durée de 9 192 631 770
périodes de la radiation correspondant á la transition entre
les deux niveaux hyperfins de l'état fondamental de l'atome de
cesium 133 (CGPM 13, 1967, Resolution 1).
(JRS: Without mention of gravitational potential)
CIPM
: At its 1997 meeting the CIPM affirmed that:
This definition refers to a caesium atom at rest
at a temperature of 0 K.
Traditionally, the Solar Day (however defined) was the fundamental unit of time, and was successively subdivided into hours, minutes, and seconds. Either 24 "equal" hours were put in each day, or 12 "equal" hours put in each daytime and another 12 "equal" hours in each night-time. Mid-day was when the Sun was in the South, which is not quite the same thing as being highest in the sky.
When clocks became accurate enough to do so, the day-length was made a constant given by the averaged apparent movement of the Sun.
These divisions were introduced, in Babylon (Mesopotamia), millennia B.C., if not earlier; try a Web search on +babylon +sexagesimal. Sixty is evenly divisible by more numbers (2, 3, 4, 5, 6, 10, 12, 15, 20, 30) than any lower number is.
Related pages include : M Fowler, WebExhibits, NIST?
So we have the minute of 60 seconds, the hour of 60 minutes, the day of 24 hours, the week of 7 named days, the fortnight of 2 weeks, the month of 28/29, 30, or 31 days, the quarter of 3 months, and the year of 12 named months.
The second is then simply 1/60 of 1/60 of 1/24, 1/86400, of a day.
The above scheme, using the averaged Sun as observed from Greenwich, gave Greenwich Mean Time.
However, the Civil Day in many places will be 23 or 25 hours (or 23.5 or 24.5 hours) long at the beginning and end of normal Summer Time.
The acronyms LMT, LCT and GCT have been used for Local Mean Time, Local Civil Time and Greenwich Civil Time. Apparently astronomers use the term GCT because, before 1925, their GMT started 12 hours late; but true Greenwich civil time in summer has been an hour ahead of GMT since 1916.
The mean solar day is not an exact multiple of the SI second. UTC is now a scale of SI seconds, with occasional adjustments. At the end of the UTC half-year, or possibly of the quarter-year or the month, the very last minute (and hence the last hour, day, month) may be slightly increased (or reduced) by the addition (or subtraction) of a single Leap Second. Note that the definition of UTC has varied, and that there is a threat of further change.
I'm told that the International Astronomical Union (IAU) still considers the day - but which day? to be the base unit of time.
In computing at least, the use of the 12-hour digital clock, with a.m. and p.m., should be rigorously avoided. Use the 24-hour clock, remembering Time Zone, Summer Time, and Leap Seconds.
Thailand uses 6- and 24- hour clocks.
See also Written Time Formats.
At least generally, these are 12-hour clocks, striking hours 1..12 and usually also each quarter of an hour.
For clocks striking thirteen, see in Sundry Quotations.
For UK definitions of Twilight, and many other things, see Whitaker's Almanac (via its index) :-
Sunset is when the refracted Sun finally disappears below the true horizon (the sea-level tangent plane). The centre is then 50' of arc (0.83°) lower.
Civil Twilight ends when the centre of the Sun is 6° below the true horizon.
Nautical Twilight ends at 12° below.
Astronomical Twilight ends at 18° below.
And conversely in the morning. See also USNO.
For current values, see Solar Data at heavens-above.
Time Zones are by definition geographical, governing Standard (Winter) Time, and do not change with the season. See Time Zone, Time offset and Time in the United States, or any good dictionary.
The International Meridian Conference [Washington, DC, USA, 1884-10-01 to 1884-11-01; Greenwich Meridian by vote on 1884-10-13 (Howse)] adopted the system of international standard time zones. There's a link for Conference details in the Greenwich 2000 millennium-faq.
Remember, when considering Time Zones, to add the effects of Summer and Winter Time.
Remember that the local standard time in some places differs from Greenwich by a non-integer number of hours (subject to change) :- -9½, -4½, -3½, +3½, +4½, +5½,+5¾, +6½, (+8¾,) +9½, +10½, +11½, +12¾, at least. Numbering the Greenwich Time Zone as 12, all integer values in [0..24] are apparently used; and the Chatham Islands are +24¾ (and add Summer Time). Tonga is +25. Part of Kiribati is +26. A date is therefore 50 hours long, overall. See List of time zones.
In the past, the time in some countries was the Solar Time (or Mean Solar Time) of the capital, with an arbitrary-looking difference from GMT. I know of nowhere which now has a difference from GMT other than an exact multiple of 15 minutes.
For the Date Line, where days begin and end, see A History of the International Date Line by R.H. van Gent.
Zones can be numbered to by giving the local standard time, on the (extendable) 24-hour clock, of noon GMT, which is generally the civil time of noon GMT in local Winter. Alternatively, by the difference from GMT in hours; it is usually obvious which is intended (but be careful about the sign of the offset).
Zones, and zones-with-offsets, can be named, and "acronyms" can be used. The terms UTC (or UT) and GMT, and no others, are internationally well-known; LCT can be used for Local Civil Time.
A Draft RFC 3339, but not the final RFC 3339, contains an "Initial List of IANA Timezone Names"; for the UK, it has "Europe/Belfast GB / Europe/London GB".
I gather that Zone 12 time (used by the UK) is called Western European Time, Zone 13 Central European Time, and Zone 14 Eastern European Time. Such terms only really make sense where the Summer Time changes are simultaneous zone-wide (or absent), as is the case in the European Union.
Standard European Time Zone Acronyms | ||||||||
---|---|---|---|---|---|---|---|---|
Number | 11 | 12 | 13 | 14 | 15 | 16 | 17 | |
Offset | -1 | 0 | +1 | +2 | +3 | +4 | +5 | |
Winter | ? | WET | CET | EET | ? | ? | ? | |
Summer | ? | WEST? | CEST | EEST? | ? | ? | ? | |
Sample | Azores | London | Berlin | Athens | Moscow | ?????? | ?????? | |
May also be seen? | ||||||||
Winter | ? | GMT | MET? | ? | ? | ? | ? | |
Summer | ? | BST | ? | ? | ? | ? | ? |
It seems reasonably certain that geographical Europe includes places in Zones 16 & 17. The EU uses only 11, 12, 13 & 14.
The acronym columns are Standard (Winter) and Daylight-Saving (Summer) designations. Upper-case acronyms are well-established; I deduced the meanings of those in lower-case here.
adt = -0300 Atlantic ast = EDT = -0400 Eastern EST = CDT = -0500 Central CST = MDT = -0600 Mountain MST = PDT = -0700 Pacific PST = ydt = -0800 Yukon yst = = -0900 hdt not needed; Yukon now not needed Hawaii hst = -1000
I've seen AKDT for Alaska and HAST for Hawaii/Aleutian.
The Chamorro Time Zone (Guam and Northern Marianas) seems to be +1000.
GDT, in RFCs 561/724, was an Americanism for BST = GMT+1.
The hourly zones or offsets are, particularly by the military, sometimes referred to by postfixed letters, or their phonetic alphabet equivalents, 25 letters can cover -12..+12, including Z for GMT/UTC.
Erroneous JavaScript and Internet interpretations of postfix zone letters are shown in my JavaScript Date and Time 0 : Date Object Information. And VBScript is different again, having only A = A.M., P = P.M.; see in my VBScript Date and Time 1.
Wikipedia Time zone - Nautical time zones links to World Time Zone Map by HMNAO, at USNO.
Alpha, Bravo, Charlie, Delta, Echo, Foxtrot, Golf, Hotel, India, Juliet, Kilo, Lima, Mike, November, Oscar, Papa, Quebec, Romeo, Sierra, Tango, Uniform, Victor, Whisky, X-ray, Yankee, Zulu.
Time designators such as BST and PST are potentially ambiguous and therefore unsafe in an international context. For example, BST is used for British Summer Time, British Standard Time, Bangkok Standard Time, Brazil Standard Time, and Bering Summer Time; EST is Eastern Standard Time, for Australia and America; IST is Irish Summer Time, Israel Standard Time and Indian Standard Time.
Note that there can be ambiguity when using postfix letters. In some contexts, 9:00A and 9:00P represent nine o'clock in the morning and in the evening, but in others, they represent 09:00 in the correspondingly lettered time zones.
Internet RFC 733 (1977), and RFC 822 (1982) section 5.2, are wrong (or, to be charitable, ambiguous) as to the "sign" of the meaning of postfix zone letters. RFC 2822 (2001) section 4.3 admits the error; it disallows the practice. MSIE4 JavaScript follows RFC 822. RFCs.
The following large countries use more than one Time Zone (within a Zone, Summer Time rules may vary). :-
China and India, however, do not.
When a older Windows PC changes Time Zone, the local time needs to be reset as well as the Zone, I hear; though perhaps not in the later versions (XP & higher?). The setting is stored in the Registry.
Some (at least) versions of Windows, including XP sp2, have a program w32tm.exe on the Path; it does various related things : see w32tm /? | more. In particular, w32tm /tz displays the current time zone settings; but the names are non-authoritative.
XP sp2, here : Winter : Prompt>w32tm /tz Time zone: Current:TIME_ZONE_ID_STANDARD Bias: 0min (UTC=LocalTime+Bias) [Standard Name:"GMT Standard Time" Bias:0min Date:(M:10 D:5 DoW:0)] [Daylight Name:"GMT Daylight Time" Bias:-60min Date:(M:3 D:5 DoW:0)] Summer : Prompt>w32tm /tz Time zone: Current:TIME_ZONE_ID_DAYLIGHT Bias: 0min (UTC=LocalTime+Bias) [Standard Name:"GMT Standard Time" Bias:0min Date:(M:10 D:5 DoW:0)] [Daylight Name:"GMT Daylight Time" Bias:-60min Date:(M:3 D:5 DoW:0)]
The means of setting the Time Zone and Offset is system-dependent.
I have a section on Time Offset Setting for News, Mail, etc.
When programming differences of times, remember the possibility of a Time Zone or Summer Time change. Remember that a computer can be set for an incorrect location; and that the system rules may be in error, and may only intended to be the current rules anyway. GMT is preferable.
To program conversion between Local Time and GMT in 32-bit Windows, see "TIME_ZONE_INFORMATION", etc., in Win32 Help. It includes the rules for Summer Time, though that is an independent concept. GetTimeZoneInformation is in the Windows unit. See my Delphi (D3 DCC32 -cc) program tz-check.pas, which includes demonstration calls to other Windows time routines.
Some systems understand a UNIX-style TZ environment variable; it is expected to be present in UNIX, and programs translated to DOS from UNIX may honour it. When it is absent, programs using it may default to foreign settings, typically those of California.
Tony Walton wrote: This is documented on the environ(5) manual page ... . A web search for "The zero-based Julian day" (in those quotes) finds many similar versions of that (h t).
It is not absolutely clear whether the second, optional offset, if present, is from local Winter Time or from GMT. Consensus is the latter.
The setting
TZ=GMT0BST-1,M3.5.0/01:00,M10.5.0/02:00
suits the UK; but some say that
TZ=GMT0BST,M3.5.0/01:00,M10.5.0/02:00
is safer. I am assured that they should have the same effect. I'd like reliable foreign samples; from away from the UK time zone, from somewhere with Summer Time at New Year, and from somewhere without Summer Time. I don't see what its TZ string should be if Newfoundland changed with the EU, though; it's too far West.
I read that another setting
TZ=GMT0BST,3,-1,0,3600,10,-1,0,7200,3600
is equivalent, apparently used by IBM in OS/2.
For permanent GMT, use TZ=GMT0.
I have a DOS utility, GAWK (ex-UNIX), which only partly understands TZ and has ISO Week Number errors.
I have JavaScript to convert a given UTC, or running current UTC, using a TZ string, to the corresponding civil time. Also my OS Time Offset Settings includes code to show the user's TZ string.
TZ allows a resolution of one second; but ISO 8601 expects UTC offsets to be multiples of a minute.
My page Date and Time Elsewhere demonstrates handling of time zones, Summer Time, and TZ in JavaScript. And Generate a User TZ String gives a draft TZ string for the user's settings.
From my general knowledge, aided by Peter Kemp's book "The Oxford Companion to Ships & the Sea" and by "Schott's Original Miscellany", the Royal Navy traditionally divides the day as listed, where the digits give the number of bells rung at the end of each successive half hour of the Watch.
Watch Ship's Time Bells Middle 00:00-04:00 1 2 3 4 5 6 7 8 Morning 04:00-08:00 1 2 3 4 5 6 7 8 Forenoon 08:00-12:00 1 2 3 4 5 6 7 8 Afternoon 12:00-16:00 1 2 3 4 5 6 7 8 First Dog 16:00-18:00 1 2 3 4 Last Dog 18:00-20:00 1 2 3 8 First 20:00-24:00 1 2 3 4 5 6 7 8
Bells are struck in pairs, odd bell last.
Sixteen Bells may be rung to mark the turn of the year.
Other systems of watches have been used.
Whenever considering a problem with date or time, bear in mind any possible significance of such things as :-
and give any relevant information.
When calculating with time, avoid as far as possible the use of multiple units, except where required for input or output; use a single unit which can be considered as of constant length for the purpose. But in administrative work, it is important that the exact stipulated definition be used, whatever it may be.
My 16-bit program longcalc, available from the Web directory programs/ as PAS, EXE, TXT, & ZIP and run at a DOS prompt, can inter alia convert between seconds and Gregorian Date. For example,
>longcalc 2001 2 3 4 5 6 #ds 1601 1 1 0 0 0 #ds sub wrt (seconds) wrt ... +12,625,646,706 seconds >longcalc 1970 1 1 0 0 0 #ds 2 31 pow add #sd ... Year: +2,038 Month: +1 Day: +19 Tue Hrs: +3 Mins: +14 Secs: +8
thus 2001-02-03 04:05:06 is 12,625,646,706 seconds from 1601-01-01 00:00:00, and 231 seconds from 1970-01-01 00:00:00 is 2038-01-19 03:14:08. The range greatly exceeds the duration of the universe, backwards and forwards; BC uses astronomer's notation in which BC 1 is 0, BC 2 is -1, ...
Program mjd_date, loc cit, has a range of ±>30000 years, in days, with more date-specific options.
JavaScript can easily do it : see in JavaScript Date and Time 2 : Demonstrations. Beware of systems non-compliantly showing Leap Seconds.
Because of Summer Time transitions, one must not assume that the hours always occur in regular succession, nor that a civil day is always the same length.
An event scheduled (in the EU) for a civil time in the early morning could be missed on the last Sunday of March, and/or done twice on the last Sunday of October.
If a unit of an hour (or less) is used to implement civil date arithmetic, or vice versa, there can be error if the transition is spanned.
It might be possible to define an "unsafe time range"; if in the UK the selected civil time is between (inclusive) 01:00 and 02:00 (02:00-03:00 in most EU, and 01:00-03:00 in NA), then there is an increased risk of error in local work. Unless extra care is taken, some scheduled operations may be better done outside those times.
Leap Seconds are less likely to affect scheduling; they are small, and most computers do not show them.
Random Number generators can be initialised from the clock; so there is a possible problem with scheduled programs.