This paper documents the Galilean satellite ephemerides designated as E5, which were delivered in support of the Galileo space mission to Jupiter. The E5 ephemerides supersede the E4 ephemerides, which were developed (Lieske 1994a) without using CCD astrometric data in order to assess the new data type. It is believed that the E5 ephemerides are better than the E3 and E4 ephemerides and they are recommended for general usage. The parameters of E5 are given in the B1950 system so that the galsat software (Lieske 1977) can be employed directly to compute coordinates in the B1950 frame, which has been adopted for the Galileo mission.
The ephemerides E2 (Lieske 1980) were developed prior to the Voyager mission and were based solely on an analysis of earth-based observations. The E2 ephemerides utilized mutual event data from 1973 (Aksnes & Franklin 1976), photographic astrometric observations from 1967-1978 (Pascu 1977 1979), and Jovian satellite eclipse timings from 1878-1974 (Pickering 1907; Pierce 1974; Lieske 1980).
Post-Voyager mission ephemeris improvements yielded ephemerides E3, which included Voyager optical navigation astrometric data and Voyager-derived physical constants (Campbell & Synnott 1985). The E3 ephemerides employed mutual event data from 1973 and 1979 (Aksnes et al. 1984), Voyager optical navigation astrometric measurements from 1979 (Synnott et al. 1982), additional photographic observations by D. Pascu from 1973-1979, and eclipse timings from 1652 to 1983 (Lieske 1986, 1987).
The initial pre-Galileo mission ephemerides were designated
E4 (Lieske 1994a) and included extended mutual event data and photographic data,
but no CCD observations, since they were still in the process of being
evaluated. The E4 ephemerides employed the previously mentioned Voyager data,
mutual event data from 1973 and 1979 corrected for phase effects by adding
to the observation time (Aksnes et al. 1986), photographic data and
Jovian eclipse timings, as well as additional mutual event astrometric
measurements from 1985 and 1991 (Aksnes et al. 1986; Franklin et al. 1991;
Kaas et
al. 1997; Descamps 1994; Goguen et al. 1988; Goguen 1994; Mallama 1992), and
additional photographic observations from Pascu (1993) covering the interval
1980-1991. Three-years' of CCD data from Flagstaff (Monet et al. 1994; Owen
1995) were evaluated, but not employed in developing the E4 ephemerides.
The E5 ephemerides represent the most current evolution of the Galilean satellite ephemerides and incorporate all of the above data types, including an evaluation the Doppler data of Ostro et al. (1992).
The 50 parameters which define the theory of motion of the Galilean satellites (Lieske 1977) could also be transformed in a manner such that the same galsat computer program can be employed to compute rectangular coordinates with their values being in the J2000 system. Documentation and an algorithm for such transformation of all galsat-related ephemerides (e.g., Lieske 1977, 1980; Arlot 1982; Vasundhara 1994) will be issued later. In the meantime the equatorial coordinates can be transformed in the following manner.
For the Galileo mission, all input quantities are in the B1950 frame and Earth
equatorial coordinates transformation from B1950 to J2000 when necessary is done
by the matrix multiplication 
where the matrix A could be taken from that recommended by IAU Commission 20
(West 1992),
with 
being the standard IAU precession matrix from B1950 to J2000
(Lieske 1979),
or A could be taken from the earlier discussion of Standish (1982),
which was developed for transforming from DE118 to DE200,
It essentially consists of a rotation
in the B1950 equatorial plane from the FK4 origin to the dynamical
equinox and then precessing from B1950 to J2000 using the IAU 1976 equatorial
precession parameters (Lieske et al. 1977).
The matrix A could also be derived from Lieske's discussion (1994b) on the
precession of orbital elements,
For the Galileo mission, the method of Standish given in Eq. (4) is employed to
precess from B1950 to J2000.
The rotation matrices Ri are the standard matrices for rotations about the x,y, or z axes for i=1,2,3:
The various matrices mentioned in Eqs. (2 (click here)), (4 (click here)) and (5 (click here)) are presented in Table 1 (click here). The maximum difference in satellite coordinates, due to the different precessional transformations, is about 1.5 km, so any of the previously mentioned matrices could be used in a practical situation.
Eq. (2): Commission 20 matrix from 
| 0.9999256794956877 | -0.0111814832204662 | -0.0004859003815359 |
| 0.0111814832391717 | 0.9999374848933135 | -0.0000271625947142 |
| 0.0048590037723143 | -0.0000271702937440 | 0.9999881946023742 |
Eq. (4): Standish matrix from 
-0
| 0.9999256791774783 | -0.0111815116768724 | -0.0048590038154553 |
| 0.0111815116959975 | 0.9999374845751042 | -0.0000271625775175 |
| 0.0048590037714450 | -0.0000271704492210 | 0.9999881946023742 |
Eq. (5): Lieske matrix from 
| 0.9999256795268940 | -0.0111810778339439 | -0.0004859930159015 |
| 0.0111810775053504 | 0.9999374894281627 | -0.0000272382503387 |
| 0.0048599309149990 | -0.0000271030297995 | 0.9999881900987267 |