\input basic2
\beginpreprint

\null\vfill

\title HST IMAGERY OF THE NON-EXPANDING, CLUMPED ``SHELL''

\title OF THE RECURRENT NOVA T PYXIDIS\/$^{a,b}$

\author Michael M.\ Shara, David R.\ Zurek, Robert E.\ Williams

\affil Space Telescope Science Institute

\affil 3700 San Martin Drive, Baltimore, MD  21218

\author Dina Prialnik

\affil Department of Geophysics and Planetary Science, Tel Aviv University

\affil Ramat Aviv 69978 Israel

\author Roberto Gilmozzi

\affil European Southern Observatory

\affil Karl-Schwarzchild-Str 2, D-85748, Garching, Germany

\author Anthony F.\ J.\ Moffat

\affil D\'epartement de Physique, Universit\'e de Montr\'eal

\affil Montr\'eal, QB  Canada

\vfill

\tobe{July 1997}{Astronomical Journal}

\recacc 7 October 1996; 16 April 1997

\vfill

{
\singlespace

$^a$Based on observations with the NASA/ESA {\it Hubble Space
Telescope}, obtained at the Space Telescope Science Institute, which
is operated by AURA, Inc., under NASA contract NAS5-26555.

$^b$Based on observations with the New Technology Telescope of the
European Southern Observatory at La Silla, Chile.

}


\endtitlepage

\abstract{
The ``shell'' of the recurrent nova T~Pyxidis has been resolved with
HST imagery into more than two thousand individual knots. Hydrogen and [NII]
emission have very similar spatial distributions. Four epochs of 
observation allow us to place an upper limit of 40~km/s 
on the systematic expansion velocity of the knots. This is in contrast to our 
1985 long slit spectra which show differential velocities of 350~km/s,
and post-outburst (1966) spectra which show velocities almost three times 
larger. The azimuthally averaged surface brightness radial distribution 
exhibits nine distinct peaks. A multiple shell model is required to 
fit the data. Our best estimate of the shell mass is approximately
$M_{sh} = 1.3\times10^{-6}$~M$_{\odot}$, with
our inability to determine the electron density
limiting any attempt to better constrain the shell mass.
A few knots are observed to fade or brighten significantly
on a timescale of months.
We outline a model in which successive eruptions of T~Pyx give rise to
collisions and shocking of successive generations of ejecta. This leads to
the clumping, emission line ratios and knot variability that we observe.
We also note that other nova ``shells" might be as highly structured as 
T~Pyxidis; and that the standard picture of uniform, monolithic nova
shells is probably a great oversimplification.}