We address the problem of network booting: Distributed processes
boot one after the other at unpredictable times in order to start some
distributed algorithm; we consider clock synchronization algorithms
in systems of n >= 3f+1 processes where at most f exhibit Byzantine
behavior. Obviously, assumptions like ``there are always at most
one third of the running processes Byzantine faulty'' do not
hold during system startup.

Using a partially synchronous model where upper and lower bounds
upon transmission and computation are unknown, we show that a
suitable modification of Srikanth & Toueg's non-authenticated
clock synchronization algorithm handles network booting and guarantees
bounded precision both during normal operation and startup. Accuracy
(clocks being within a linear envelope of real-time) is only guaranteed,
when sufficiently many correct processes are eventually up and running.