[LEAPSECS] the big artillery

[Dennis Ferguson]

On 30 Oct, 2014, at 12:12 , Richard Clark <rclark at noao.edu> wrote:
> Well, for historical and archival purposes Julian date nearly always means
> traditional days, as in solar days. But for astronomical uses a fixed
> unit, the apocryphal atomic day is implied. This means needing to know
> delta T if you need to relate it back to a civil date or time.
> 
> The term 'day' has an awful lot of linguistic baggage that clearly
> implies that the solar day is meant. But now the use of 'day' can be
> at the speaker's and listener's risk.
> 
> The minute, hour, day, year... these are not SI units. We need to
> start considering it sloppy to use them as if they are.
> 
> Do we mean 'atomic day'? If so we need to:
> 1. say so, and
> 2. make it official by defining, rather than just implying one.
> 
> Perhaps hectosecond would be better. At least it doesn't invite
> confusion.
> 
> Yeah, and now to convince anyone to do this.

I agree with that.

While it is true, though, that the minute/hour/day are not SI units
they are accepted by the CIPM for use with the SI with dimensions
of time using their traditional (dating back to Ptolemy?) defined
relationships to each other when the second used is the SI version.
Table 6 here

    http://physics.nist.gov/Pubs/SP330/sp330.pdf

lists the CIPM definition, to which TT as a JD seems to conform.
Solar days/hours/minutes/seconds don't conform, but since their
dimensions are now rotational angle rather than time this is a
use of the units which is off label with respect to SI even if
it is a traditional and original use of units with those names.

I guess the point is that while "day", like many traditional
units, is ambiguous and in need of a qualifier to know exactly
which kind is being referred to, a definition of "day" as a
standard time unit which is 86400 SI seconds long already exists
and is in use.  This requires no new invention.

Dennis Ferguson