Blueshifted AGN absorption lines indicate outflowing material from the central engine. Many repeat
spectral observations of AGN have observed variability in the shape and strength of these lines, however
there is no scientific consensus regarding the dominant mechanism which drives this variability. The
main purpose of this thesis is to investigate this mechanism in a sample of BALQSOs and a Seyfert 1
galaxy.
The BALQSO sample consists of 50 quasars, in which we test the dependence of broad absorption line
variability on several variables. To do this a novel technique is used (Non-negative matrix factorisation)
to reconstruct the entire emission profile, allowing accurate measurement of variability even in cases of
absorption line overlap with the corresponding broad emission line. No significant correlations between
variability and any of the investigated properties are found. The lack of luminosity correlation suggests
ionisation changes do not play a dominant role in the variability of the BALQSO population.
The most variable BALQSO in the 50 quasar sample, SDSS J113831.4+351725.3, shows extraordinary
BAL variability, with one C IV BAL lying outside the random walk model which accurately predicted
the BAL behaviour in a large BAL study. Co-ordinated BAL changes in this object, despite being separated
by thousands of km s−1, together with results of photoionisation simulations and the behaviour
of the continuum, strongly suggest that the variability mechanism is ionisation fraction changes driven
by variations in the input ionising continuum.
The Seyfert 1 galaxy NGC 4151 is noted for its continuum variability, however its absorption line
variability is less well studied. Examination of the variability of metastable helium absorption lines,
which are sensitive to high column densities, suggests a crossing-clouds scenario in which changes to
the line-of-sight covering fraction occur, thereby driving spectral changes. Photoionisation calculations
suggest a mass outflow rate which is too small to result in galaxy feedback effects.