Hacker News Reader

Comment by droidist2 | Hacker News Reader

> Interestingly, in order to keep atomic time in agreement with astronomical time, leap seconds occasionally must be added to UTC. Thus, not all minutes contain 60 seconds. A few rare minutes, occurring at a rate of about eight per decade, actually contain 61.

So the second should be a tiny bit longer? Like 1.00000015 times longer?

View on HN

Leap seconds can be positive or negative (never happened) and are irregularly announced up to two times a year.

The wikipedia article on leap seconds[1] is worth reading and has this[2] great figure showing the variations of day length over the years due to various astronomical factors. It also shows how these small deviations slowly desync UT1 and UTC (red line), and that this issue cannot be solved by a linear scaling.

[1] https://en.wikipedia.org/wiki/Leap_second

[2] https://upload.wikimedia.org/wikipedia/commons/thumb/5/5b/De...

Surprisingly, I've found explanations that it's due to variation in the year and day. Remember, the year is based on the physical phenomena of goin'round the sun, and so we have the issue of theory meeting practice, where our ideal of a year fails to model the truly complex dynamics of the actual earth's orbit. There are errors being introduced due to the moon slowing our spin, other planets distorting our orbit, and other such subtle and difficult-to-account-for cumulative errors from ideal steady-state.

So even of the second was defined as slightly longer, we'd still have to issue corrective leap seconds.

I meant to look into this particular thing more closely last time I read about leap seconds here, so I'm glad for the reminder. If anyone has actual details and references about these effects, I'd be much obliged.

I believe it is the year that's off a bit in relation to the day.

Reply on news.ycombinator.com