This is a very rapidly changing field of research. The advent of the
10m-class optical telescopes such as Keck, soon to
be joined by
others, has revolutionised our understanding of galaxy evolution in
recent years and much still can be done in the optical and near-infrared:
the recent discoveries of Lyman 
emission from a few galaxies at
z = 3-6
(Dey et al.
1998,
Hu et al.
1998),
shows that not all galaxies at this epoch
are dusty, which provides fresh hope for the discovery of PGs when the
new generation of even
larger near-IR arrays becomes available on the largest telescopes. Certainly
the US space agency NASA
recognises the importance of continuing PG searches: over the next
two decades it has dedicated itself to the ``cosmic
origins'' programme, designed to answer such questions as ``how did the first
galaxies form?''This will require a succession of sophisticated telescopes,
each building on the results of previous missions augmented with ground-based
observations. One such project is the New Generation Space Telescope -
a 4m (or possibly 8m) descendant of the incumbent Hubble Space
Telescope, designed for
diffraction-limited imaging over 1 deg2. With such
capability it will be possible to observe the individual supernovae explosions
resulting from massive bursts of star formation happening at almost any
epoch.
Undoubtedly, far-infrared and submillimetre observations will have future
role to play in the unfolding story: the Planck satellite mission (named
after the German physicist Max Planck), is
part of the European Space
Agency's Horizon 2000
Scientific Programme. Due to be launched in 
2005, Planck will
survey the whole sky at millimetre wavelengths with
unprecedented sensitivity and angular resolution. Designed primarilly
to map fluctuations in the cosmic microwave background, Planck has the
capability to detect small fluctuations in the far-infrared background
tentatively detected by COBE, thereby distinguishing between rival
theories predicting different epochs of
galaxy formation. Perhaps the most exciting prospect on the timescale of
10 yrs or so, is the development of
millimetre array
imagers, capable of both high spatial and spectral resolution imaging
over the whole sub-millimetre waveband. For example, the National Radio
Astronomical Observatories millimetre array, proposed to the US National
Science Foundation will provide a spatial resolution of 10 milliarcsec
between 10 mm - 350 µm and a sensitivity capable of detecting
the dust emission from a bright star-forming galaxy with luminosity
> 1011 L
to z = 20.
ACKNOWLEDGEMENTS
I would like to thank Gordon Rogers for the spectral energy distribution plots and Doug Burke for useful discussions.