How do clocks keep up with time?

  1. Excuse me, if i posted on a wrong forum.

    I heard this quite a time ago, that one day on Earth is not exactly 24 hours. And i'm pretty sure some unit charts on the back of composition books will say one day is 23.9 hours or something like that.

    My fantasy about how we defined the length of a day is that, in the first place, somebody divided the length of a day as exactly 24 hours, then made smaller divisions with minutes and seconds. I assume the rotational speed would not change the 60 seconds-60 minutes-24 hours scheme.

    In case of analog clocks, we can tweak it so that the hands move just a little bit faster to make them look like it always takes exactly 24 hours for a day. For digital clocks, it may be more "accurate" to tweak it so that at 24th hour, the day ends at 56 min and 4 sec and move on to 12:00:00 AM for the next day.

    I understand that without regular calibration, clocks go a little faster or slower after few months or years.

    I mean, whatever mechanisms are used to "conceal the loss" of few mins everyday, ppl can have an impression that one day on Earth is always exactly 24 hours (maybe no matter what).

    So the questions are,
    (1) Are there mechanisms for clocks, especially analog clocks, to tackle the issue of less-than-24-hours-a-day?
    (2) Why do ppl keep communicating (for example, "the deadline is 11:59:59 PM local time") as if the "loss of 4 mins is not a great deal for errors so we just slow the clocks in the background just a little bit so nobody will notice the 4-min difference"?
  2. jcsd
  3. Actually, the hour is almost exactly one twenty-fourth of the mean solar day (which is what we normally mean by the word "day").

    Due to a very small difference between the hour and the solar hour, there are occasionally "leap-seconds" added (or, theoretically, subtracted, but this has never happened) to the UTC day to keep it synchronized with the sun. Everyone else just sets their clocks relative to UTC.

    If you heard that the day is 4 minutes less than 24 hours, this probably comes from the fact that the sidereal day is (approximately) 23 hours 56 minutes long. The sidereal day is (an approximation to) the time that it takes for the Earth to rotate once relative to the stars. There are 366.24 sidereal days in a year! It is definitely not our normal concept of "day".
  4. Although the apparent solar day varies up to -21 to +29 seconds from the average. This is caused by the earths having an elliptic orbit and an axis of rotation that is not perpendicular to the plane of the orbit.
  5. Um, i forgot to point out something.

    I should have mentioned that at present days, the length of a second is standardized by the properties of a cesium atom.

    So, what this tickles me is, in the early history, did the length of a second defined by the speed of Earth's rotation? Meaning that if Earth were to arbitrarily rotate faster, (i'm talking pure science fiction here) the length of a second would be "shorter" and vice versa. If this was the case then, maybe this 24hr-60min-60sec scheme was set accordingly, then quite recently, a new standard for one second was established and that's why the new measurement of the length of one day turns out to be "a little shorter" than 24 hours.

    I'm still curious at how humans cope with this phenomenon. Are the clocks really tweaked in a way that display full 24-hour scheme, even though the cesium atom would give different information? (For example, the atom would have counted 86300 seconds, but they are programmed to display, perhaps 0.1% longer for each second) Are we so used to the 24-hour scheme, and adapting to this "micromanagement" efforts to be precise and concise just seem pointless?
  6. jim mcnamara

    jim mcnamara 1,565
    Science Advisor
    Gold Member

    Time measurement has not always been based on atomic clocks. The US Naval Observatory (WWV) used to broadcast time nation-wide. We used to use WWV to set computer times before the internet.

    Before that there was train time, and civil time, based on local noon. Astronomers at local institutions were usually the source of that data. Train time was fixed to some civil point. Each town either used train time or local time.

    Before then, and to this day, have Churches kept calendars in order to observe holidays. During the Middle Ages in Europe, local monasteries or churches rang bells for certain times, like matins. Later on church towers with hourly bells and civil timekeeping towers (Big Ben)became more common. Think of Westminster chimes to tell the time.

    Basically the ideas we use for timekeeping have been around for a very long time. It is their implementation and out ability to measure time periods that has improved amazingly in the past 400 years. Dionysius Exiguus in ~576AD got the early church to accept what became the final Julian calendar, and then it was re-mandated by Pope Gregory in 1582. This is the current Western Gregorian Calendar. Astronomers use Julian Days as days elapsed since a certain time:

    Code (Text):
    Noon Jan 1 -4712   # 4712BCE
    This time/date system evolved and used GMT as the base for the day changing noon. The observatory at Greenwich England reported the data.

    It was always time - Time again -- noon that astronomers can clearly report on and see. Calendars, time, and astronomy have LONG been tightly linked.

    You may want to rethink some of your assumptions.

    See: (Julian Numbers) for starters.
    (babylonian astonomy)

    Google for Scaliger, Lunar Calendar, or Sothic Calendar to get an idea for how long people having been striving to keep track of dates, and later times.
    Last edited: Aug 29, 2013
  7. SteamKing

    SteamKing 11,008
    Staff Emeritus
    Science Advisor
    Homework Helper

    Humans are responsible for keeping track of time, not clocks. Clocks are instruments which approximate the passage of time by mechanical means and therefore are subject to some error in operation.

    Julian Day #1 actually began at noon on Jan. 1, 4713 BC according to the current calendar:

    Remember, there was no Year 0 in the Gregorian Calendar.
  8. Nugatory

    Staff: Mentor

    There's no problem saying that a day is equal to exactly 86400 seconds (24 hours times 60 minutes in an hour times 60 seconds in a minute) using the cesium-atom definition of a second. There's also no problem building clocks that follow that standard, meaning that they'll count 24 hours passed in the same time that the cesium atom would count 86400 seconds.

    However the earth does not make a full revolution on its axis in exactly 86400 seconds (again, using the cesium-atom definition - that's what a second is). This, we have a problem if we expect noon to pretty much fall exactly between sunrise and sunset every day, as every rotation would increase the difference between noon on our clocks and the midpoint of the day.

    But we don't play games with the speed of clocks to correct for this. Instead we insert a "leap second" (google for that phrase!) every once in a while, just to keep our clocks lined up with the earth's night/day cycle. An isolated clock like my wristwatch is suddenly one second off when a leap second comes by and I don't even notice. A cell phone or an internet-connected computer is generally getting its time setting from a server somewhere and just adjusts automatically for the leap second when the server says to; again, I probably don't even notice.

    The last leap second was sometime in the summer of 2012 (and I would bet that you didn't notice it either). It's not an accident that they don't happen very often - the cesium atom definition of the second was chosen so that the average day would be as close as possible to 86400 seconds.

    BTW, the word "average" in the previous sentence is important- see glappkaeft's post above.
  9. I think I know what the issue is here. 23 hours and 56 minutes is approximately a sidereal day. That's the time it takes the Earth to rotate 360 degrees, so that the stars are in the same position. But most people are more interested in the sun's position than in the stars. Earth travels around the sun in 365 days or so, so while Earth rotates once, it's also moving about 1 degree farther in its orbit, meaning that to line up with the sun, it has to turn one degree more, which takes another four minutes. (This is also why the constellations appear to start a bit farther West each night. It's the sunset happening four minutes later in sidereal time.)

    So a 24-hour clock is just right for people on Earth and there is no need to tweak. I mean, aside from Daylight Saving Time, dead batteries and the slight imperfections of clocks of course.
  10. Baluncore

    Baluncore 3,305
    Science Advisor

    The Earth does speed up and slow down very slightly. The rate of rotation is continuously effected by variations in the distribution of the atmosphere, ocean water and the ice caps.
  11. There are some interesting insights regarding historical time corrections here:

    In the modern world, time must be recorded and distributed very accurately internationally. Years ago when I worked at AT&T they had a separate time 'synchronization network' connecting every communications office in the country. In fact there were even diversely routed back ups in case the primary synch network failed. This provided very accurate universal digital time based on an atomic clock operated by the government [Maybe it was the Navy lab already mentioned, I forget] so that every office had the same time for encoding digital transmission packets...voice and digital signals. This even accounted for signal transmission times coast to coast.

    When this thing went down, that is failed for some reason, there was fear the nationwide network would become unsynchronized...because each office began free running, usually only for a few minutes, meaning everybody was running on 'separate clocks'. If that lasted too long, digital packet timing information might be unrecognized... meaning voice or digital information could be lost because different offices could not recognize the other's clock time.

    Synchronized timing of a similar sort is utilized currently for the satellite based GPS network. As you may know satellite time and earth station time progress at slightly different rates, and all portions of the system must also be corrected for transmission times, analogous to the AT&T terrestrial network I described, so every station has a reference clock time.
  12. D H

    Staff: Mentor

    The cesium clock definition of the second was chosen so that this new definition of the second was consistent with the prior definition of the second. The previous definition had already divorced the second from the daily rotation of the Earth, defining the second instead as a fraction of a year. That older definition was based on observations from 1750 to 1892. Our 86,400 second long day essentially reflects how long a mean solar day was in 1821. We have to add leap seconds occasionally because the Earth's rotation rate is gradually slowing down. The tides make the Earth transfer rotational angular momentum to the Moon's orbit. A mean solar day is currently about 86400.002 seconds long.
  13. Electric clocks that you plug into the wall are "self calibrated" by their dependence on the AC cycles provided by the service outlet and those cycle counts are adjusted continuously by the electric service utility provider to keep on spec.

    Modern household, school, and business clocks that are powered by the electric service utility use the 60Hz (or in some countries 50Hz, etc) of the service outlet to advance their movements.

    Amazingly, the utility service actually counts and controls every cycle; if during a period of service the count lags or leads even one cycle, the utility makes a correction to account for it - in the long run the number of cycles per second is maintained to their spec Hz.

    They do this in part because these clocks depend on the Hz to keep time. This allows all of these clocks within the service area powered by the utility to maintain virtually absolute synchrony if the clocks are set precisely. Even if the clocks are not all set precisely, they will all take the same time to move one minute, one hour, etc. because the AC cycles are in synchrony and the clocks are using the AC cycles to operate their advancement. The clocks will not gain or lose time with respect to each other.

    Off course the utility references the national standard clocks to maintain their cycle count, so all those ordinary electric clocks on the walls of all those homes, schools, and businesses can keep very accurate synchronized time indefinitely without user adjustment (except after a power interruption).
  14. Drakkith

    Staff: Mentor

  15. SteamKing

    SteamKing 11,008
    Staff Emeritus
    Science Advisor
    Homework Helper

    When I worked aboard ship, electric power was of course generated aboard, but the cycles weren't regulated too accurately. As a result, the electric clocks would all start to run slow after being reset to the correct time (which was done as we crossed into different time zones during the voyage).
  16. Well, what i meant by "pure science fiction" would have to be 20 to 100 times faster than what earth does today... :)
  17. Baluncore

    Baluncore 3,305
    Science Advisor

    Yes, I realise it was “pure science fiction”, but the seasonal weather factoid is important to astronomical observations, and this is a “physics facts” forum.

    An interesting "purely science fact" is evidenced by the daily and annual growth rings in fossil corals. Earth rotation in the Cambrian was faster than now. IIRC there were about 420 days in a year. (No I am not 600Ma old, though it seems like that long since I studied geology). 420 days implies only 20.85 of our hours per day back then.

    So if anything, the Earth is slowing down. Over the last 600Ma the Earth day has lengthened by a total of 3.15 hours, which works out, as a (linear) accumulation, at about 18.9 micro seconds per year. In (1/18.9us) = 53,000 years we will have to insert one leap second every year to compensate.

    We have only wandered this planet for about 60,000 years, and have been recording time precisely for less than half a century. Since we started playing with atomic clocks the year has been extended by less than 1ms, which is well less than the variation due to atmospheric distribution.
Know someone interested in this topic? Share this thead via email, Google+, Twitter, or Facebook

Have something to add?

Draft saved Draft deleted