Starting with 2.2, the HYCOM archive files no longer include density.  This 
is because T&S define density via the equation of state.  However, this means
that the archive file must include information about the actual equation of
state.  So "k  dens" for montg1 (the first record in the archive file) are
actually sigver and thbase, where sigver is a small integer identifying the
equation of state (taken from a parameter defined in stmt_fns.h).

So far we have (odd for sigma-0, even for sigma-2, >=40 for sigma4):

sigver =  1  ---  7-term sigma-0
sigver =  2  ---  7-term sigma-2
sigver =  3  ---  9-term sigma-0
sigver =  4  ---  9-term sigma-2
sigver =  5  --- 17-term sigma-0
sigver =  6  --- 17-term sigma-2
sigver = 46  --- 17-term sigma-4
sigver =  7  --- 12-term sigma-0
sigver =  8  --- 12-term sigma-2
sigver = 48  --- 12-term sigma-4

The 17-term and 12-term versions are rational function approximations, and
are equal to the associated 25-term and 18-term sigloc evaluated at the
fixed reference pressure (0 or 2000 or 4000 dbar).  Note that the 17-term 
version does not have a closed form expression for tofsig or sofsig, and 
so can't be used if potential density is a prognostic variable.  In 
diagnostic programs, the 17-term tofsig and sofsig are calculated using 
Newton iterations from a 12-term 1st guess.