On Wed, 29 Feb 2012 15:43:02 -0500, John Gilmore wrote: >Paul Gilmartin writes: > ><begin extract> >STCKE is notionally closer to TAI than to UTC in that TAI and STCKE >are continuous timescales and UTC is discontinous. TAI and STCKE both >embody the notion of (micro)seconds since an epoch; UTC is specified >in terms of yyyy mm dd hh mm ss.fraction with minutes varying in >length as leap seconds occur. ></end extract> > >Note quite. This formulation is plausible by analogy with the notion >that the Gregorian Month of February, normally comprised of 28 days, >is comprised of 29 days in leap years. > >Leap seconds, however, are inserted into UTC by the BIPM upon the >recommendation of the IERS (Earth Rotation and Reference Systems >Service); and they are conceptually and by definition >extracalendrical. Neither 1) the last minute in June or the first >minute in July nor 2) the last minute in December or the first minute >in the subsequent January is lengthened when a leap second is inserted >between them. [This decision was taken advisedly. There are a number >of calendars---The Hebrew religious one is the obvious example---that >make no use of minutes and/or seconds.] > And here, I'll disagree with you. The embolismic second is 23:59:60 of the previous day. I've seen it flash by with a telnet connection to NIST. (Or maybe they just had to call it something.) This seems to me no more extracalary than February 29. But it's all highly academic.
By the way, the embolismic day in bissextile years is February 24, the sixth day before the kalends of March. (Spellcheck hates us.) -- gil ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
