I was leafing through the book "The Computer" by Mark Frauenfelder. In there, I
saw an advertisement by UNIVAC that noted that their computer was the one
powerful enough to be entrusted with the job of finding a lost moon of the planet
Jupiter!

Well, I knew that some asteroids had been lost and then recovered when a
re-discovery and orbit computations led to it being determined that the new
asteroid was the same as the old one. But this was the first time I heard of this
happening with a planetary satellite.

I Googled the story, and I did manage to find it.

The story may have been less dramatic than the disappearance of some asteroids,
but saying the moon was "lost" until the UNIVAC was put to work wasn't that
inaccurate either.

An article published in one of Univac's own magazines, Systems Magazine, was
reprinted in Computers and Automation, and the issue involved is available on Al
Kossow's Bitsavers site; it's the November 1955 issue. (Also available from the Internet Archive.)

The moon in question was Jupiter VIII, now named Pasiphaë.

This moon has an orbit fairly distant from Jupiter, and thus the Sun's gravity is important enough to make its orbit very much an example of a lunar-type orbit, a case of the three-body problem that is not well-approximated by a closed-form solution.

So back in 1938, using such automated assistance as was then available, one Herb Grosch - later famous as a columnist for Computerworld, and as a relatively nonconformist employee of IBM - in his 1942 Ph.D. thesis performed integrations of the orbit of Pasiphaë which ran until 1941.

Herb Grosch is perhaps best known for "Grosch's Law": that the power of a
computer is proportional to the square of its price. That made sense back then
when a computer had to be very big to use various techniques that made
computers more efficient - longer registers, hardware floating-point, pipelines
- of course, today, these techniques are seen even in single-chip
microprocessors.

Paul Herget, working at the observatory of the University of Cincinnati, was therefore involved with the calculation of asteroid positions for the IAU.. Dr. S. B. Nicholson, the discoverer of several of Jupiter's moons, sought to photograph Jupiter VIII, and wanted its current position.

And so Dr. Herget took on extending Herb Grosch's integrations, using time (and
programming assistance) offered by Remington Rand on their Univac computer, to
1980, as a pilot project to test the feasibility of using computers for
asteroid ephemerides. It was a success; the calculated positions were "accurate
to one minute of arc".

In my Googling, I found other interesting information as well.

Wikipedia had preserved the older, unofficial, names for several of Jupiter's satellites that people had used occasionally from 1955 to 1975.

While the names of the Galilean satellites were proposed shortly after Galileo
discovered them, it wasn't until much later that those names were routinely
used; for a long time, just calling those moons Jupiter I through IV was more
common.

Camille Flammarion was responsible for the unofficial name Amalthea for Jupiter
V; that one was ultimately adopted.

And when Charles Kowal discovered Jupiter XIII, he proposed the name Leda to
the IAU, and that became the official name; and since then, the naming history
of Jupiter's moons has been smooth.

But Jupiter VI through XII were originally named Hestia, Hera, Poseidon, Hades,
Demeter, Pan, and Adrastea, in order.

And as these names were felt perhaps "too grandiose", their official names
instead are now Himalia, Elara, Pasiphaë, Sinope, Lysithea, Carme, and Ananke.

So even before Pluto, there was cause to view the IAU as a party-pooper!

John Savard