Explanatory notes:

The electron moments included in this data set are derived from fitting a Kappa
model to energy spectra of solar wind electron distributions measured by the
Wind/SWE electron instrument (see Ogilvie et al., "SWE, a comprehensive plasma
instrument for the Wind spacecraft", Space Sci. Rev., 71, 55, 1955).  Moments
parameters are computed from 9s measurements which are usually separated by one
or more 3s spin-periods.  These quantities are reliable and citable with
caution, meaning that the PI advises that the user should discuss their
interpretations with a member of the SWE science team before publishing.

The following comments are intended to aid in the use and interpretation of the
prime quantities of this data set, the electron density and temperature.  We
begin our analysis by assuming that the solar wind plasma being observed has
electron density, N [#/cc], and electron temperature, T [K]; the goal being to
indirectly determine these.  The (13) energy channels over which observations
are made are: E = 19.34, 38.68, 58.03, 77.37, 96.71, 116.1, 193.4, 290.1, 425.5,
580.3, 773.7, 1006., and 1238. eV; f(E) [#/{cc*(cm/s)^3}] is obtained for each
E--constituting an 'energy spectrum'.  The unitless parameter, Kappa, varies
over the open--endpoints excluded--interval: (1.6, 6), with a typical value of
3.  W [eV], the average thermal energy of an electron in the observed plasma, is
defined as: {[Kappa - (3/2)]/Kappa}*kT, where k is Boltzmann's constant [eV/K].
The "Kappa model" is defined as: f(E;A,s,eta) = A*[1 + (E/eta)]^-s, where A =
{N/[2*pi*(eta/m)]^(3/2)}*[Gamma(s)/Gamma(s-{3/2})], s = Kappa+1, and eta [eV] =
Kappa*W.  Note that A has the same units as f(E), m is the mass of an electron,
and 'Gamma' refers to the "gamma function".  If--by selection of A, s, and eta--
we successfully "fit" f(E) to our (E,f) measurements, we may "unpack" (solve
algebraicly for) N and T.  Note that a constrained nonlinear optimization
algorithm is used in the fitting process, with each energy channel's counts
being weighted according to the standard error assumed for a Poisson-distributed
process.  The "quality" of each fit is indicated as a unitless value, 0 <= Q <=
1, such that Q will be near 0 for a poor quality fit, and near 1 for a good
quality fit.  The data set reported here contains: N, T, W, Kappa, A, and Q (as
described above).