We have used the IUE to the limit of its capability (similar to 10(-15
) erg cm(-2) s(-1) Angstrom(-1)) and present results based on long ter
m monitoring in the UV of the Herbig-Haro object HH 29, which is dynam
ically coupled to the outflow activity from the deeply embedded low ma
ss stellar object IRS 5 in the L 1551 dark cloud. The eight years of I
UE observations confirm the degree of variability of the object, origi
nally discovered by Cameron & Liseau (1990; A&A 240, 409), both what c
oncerns the amplitudes (factor of two) and short time scales (less tha
n 0.5 yr). We have now also found declines in brightness with these sh
ort time scales, implying local particle densities clearly in excess o
f 10(4) cm(-3). The variations of the shortwave continuum (similar to
1200-1950 Angstrom) and of the high ionization species follow a simila
r pattern, whereas the intensity variations of forbidden lines from lo
w ionization species appear anti-correlated. Such behaviour is consist
ent with HH 29 changing its degree of excitation with time, probably b
ecause of multiple shocks in the object. We argue that the Mg II h&k l
ines are optically thick which would explain why they are observed not
to vary. From this we estimate that the true variability time scale o
f HH 29 is of the order of weeks (10(6) s) rather than that determined
by the observing frequency, which is several months (10(7) s). The sl
ope of the very blue shortwave continuum varies in time as well, which
we interpret to be caused by changing temperatures as a consequence o
f the different shock waves passing through the object. Combining the
IUE data with simultaneous ground based observations leads us to const
ruct a two-phase model for HH 29 from which we derive the physical par
ameters of the object. In addition to a conventional component (10(4)
K and 10(3) cm(-3)), a hot component (several times 10(4) up to more t
han 10(5) K) having average densities at least as high as 10(6) cm(-3)
is required to reconcile with the observations. The volume filling fa
ctor of this gas is consequently small (on the order of 0.1-1%). From
comparisons with the high excitation objects HH 112 we infer that the
multi-phase conditions characterizing HH 29 probably also apply to oth
er, less systematically observed HH objects. From the UV luminosity ge
nerated by the shocks in HH 29 (0.5 L.) we infer a lower limit to the
rate at which the central source IRS 5 loses mass (> 2 10(-5) M. yr(-1
)). This limit on the mass loss rate is considerably larger than any c
onceivable mass accretion rate for the stellar object (probably signif
icantly less than 10(-5) M. yr(-1)). Given also the observed variabili
ty of HH 29 we argue that the flow from IRS 5 is probably neither homo
geneous nor steady in time. This could potentially reduce the mass los
s rate otherwise needed to account for the observed radiative losses i
n the L 1551 flow.