To:   GRO pulsar people
From: J. H. Taylor
Date: March 13, 1991

Greetings from Australia!  As many of you know, I'm here at the
Australia Telescope National Facility for several months, and
one of the projects I'm working on is coordination of the radio
observers' efforts to make pulsar observing possible with GRO.

After much discussion of the issues, Dick Manchester and I
have settled on a specific format for data files containing
the information you'll need.  Most of the essential information
will be in a file called "psrtime.dat", available by anonymous ftp
from puppsr.princeton.edu (or Internet address 128.112.128.165).
After you login with username "anonymous" and using anything for
a password, type "cd gro" to change to the GRO directory.
Then you can "get psrtime.dat" to download the main data file.
A copy of this message (or suitably updated instructions) will
be available as the file "read.me", and the file containing binary
pulsar data is "psrbin.dat".

A short dummy version of the main file looks like this (you'll have
to list it on a wide or "landscape mode" (120 column) printer to
see the full width):

 PSR B     RA(J2000)   DEC(J2000)   MJD1  MJD2    t0geo(MJD)         
f0(s^-1)      f1(s^-2)     f2(s^-3)  RMS O B
-------------------------------------------------------------------------
----------------------------------------
0531+21  05 34 31.974  22 00 52.05 48282 48316 48299.000000134  
29.9516010684378 -3.77726D-10   0.00D+00  5.5 P 
0655+64  07 00 37.811  64 18 11.26 44546 47829 46187.000002101   
5.1106207921632 -1.76088D-17   0.00D+00  0.7 P *
0833-45  08 35 20.680 -45 10 35.70 48256 48320 48288.000000968  
11.1989616062225 -1.55846D-11   1.08D-21  0.5 P 
1855+09  18 57 36.394  09 43 17.33 46437 48239 47338.000000045 
186.4940816803072 -6.20316D-16   0.00D+00  0.2 P *

The column headings should be reasonably self-explanatory:

PSR	Pulsar name in 1950 coordinates (for consistency with prior usage)
RA	Right Ascension in J2000 coordinates (hh mm ss.sss)
DEC	Declination in J2000 coordinates (sdd mm ss.ss)
MJD1,2	First and last dates for valid parameters (MJD)
t0geo	Infinite-frequency geocentric UTC arrival time of a pulse (MJD)
	Note: the integer part of t0geo is the barycentric (TDB) epoch
	of RA, DEC, f0, f1,and f2
f0	Pulsar rotation frequency (s**(-1))
f1	First derivative of pulsar frequency (s**(-2))
f2	Second derivative of pulsar frequency (s**(-3))
RMS	Root-mean-square radio timing residual, in milliperiods
O	Observer code
B	Blank for single pulsars, "*" for binaries.

The "Observer codes" have the following meanings and contact persons:
A	Australia Telescope National Facility (Dick Manchester)
B	Bologna (Nichi D'Amico)
C	Cornell/Carleton (Jim Cordes, Joel Weisberg)
J	Jodrell Bank (Andrew Lyne)
P	Princeton (Joe Taylor)

For binary pulsars, the orbital parameters will be put in a file
called "psrbin.dat".  A short dummy version of the file looks like this:

 PSR B           pb         a1*sin(i)      e          t0           omega 
   omdot    gamma     pbdot   O
-------------------------------------------------------------------------
-------------------------------
0655+64      88877.061781   4.1255679 0.00000000 46386.41030392 
180.000000 0.00000 0.000000   0.00D+00 P
1855+09    1065067.590793   9.2307810 0.00002157 47529.90001480 
276.493717 0.00000 0.000000   0.00D+00 P

where the parameter definitions are (for details, see Taylor and 
Weisberg,=20
Astrophys. J. 345, 434, 1989):

PSR	Pulsar name
pb	Orbital period (s)
a1sini	Projected semi-major axis (light seconds)
e	orbital eccentricity
t0	Barycentric time (TDB scale) of periastron (MJD)
omega	Longitude of periastron (degrees)
omdot	First derivative of omdot (degrees per Julian year)
gamma	Time-dilation and gravitational redshift parameter (s)
pbdot	First derivative of pb
O	Observer code

Many binary pulsars will not require the last three (relativistic)
parameters.


We are not including uncertainties in the parameter file, and not all
of the listed digits are necessarily significant. 
Instead, we list MJD1 and MJD2, the starting and ending MJDs on which
the parameters are "guaranteed" to be valid, and RMS, which should
be a conservative estimate of the uncertainty in estimating pulsar
phase between dates MJD1 and MJD2.  In many (or even most) cases, good
phase predictions will be possible well outside the specified range
of dates.  For example, the advertised range of validity for
PSR 0655+64 in the table above extends from November 1980 to
October 1989, but in practice it will almost certainly be good until
1995 or later.  On the other hand, noisy pulsars like the Crab and
Vela require nearly concurrent data for good phase "prediction".

Note that there may in multiple entries for a given source in the
files.  Users will have to select the ones they want to use according
to range of dates -- and for pulsars being observed by more than one
of the radio observing groups, there may even be overlapping ranges
of dates.  A suitable choice might then be made on the basis of the
smallest listed RMS.

A few additional notes.  As outlined in my e-mail to Mel Ulmer
of January 28 1991, we will always try to provide enough information
to permit phase prediction to an accuracy of 0.01 cycles, i.e.,
10 milliperiods.  For many pulsars this will require just one set
of parameters and just one derivative, namely f1.  Noisy pulsars
will be better characterized with two derivatives (f1 and f2), and
where this is not good enough we will split the data into smaller
segments and provide additional lines in the table with distinct
starting and ending dates.

The sample tables reproduced above contain real data.  Since launch
is imminent, and you'll probably want to observe the Crab and Vela
for calibration purposes, we'll try to keep their parameters as
up-to-date as possible, and we'll also add more pulsars soon.  You can
fetch the files (as outlined above) and try them out as soon as you
wish.

One final warning: DO NOT trust the geocentric pulse arrival times
yet!  More particularly, do not use them to compare gamma ray
phases with radio phases, without direct contact with the radio
observer(s).  Many subtle issues arise in getting such an alignment
correct, and specific cases will need to be discussed in detail.
More generally, when we get to the point of analyzing real signals
and drwaing scientific conclusions from them, I would urge GRO
investigators to contact the radio observers for more detailed
and carefully calibrated ephemerides, with error bars.

Finally, a word for programmers.  The Fortran statements used to
write out the file entries are listed below:

Main file:

	  write(33,1120) psrname,irh,irm,rsec,decsgn,idd,idm,dsec,
     +      mjd1,mjd2,t0geo,f0,f1,f2,rmsmp,obsflag,binflag
1120	  format(a8,2i3.2,f7.3,1x,a1,i2.2,i3.2,f6.2,2i6,f16.9,
     +      f18.13,1p,d13.5,d11.2,0p,f5.1,2(1x,a1))

Binary pulsar file:

	  write(34,1100) psrname,pb,a1,e,t0mjd,omz,omdot,gamma,pbdot,
     +      obsflag
1100	  format(a8,f17.6,f12.7,f11.8,f15.8,f11.6,f8.5,f9.6,
     +      1p,d11.2,0p,1x,a1)