The various sources of noise are combined in Table I, where
the cirrus emission has been taken at the average for high latitude
sky. The behavior across the far infrared band progresses rapidly
toward increasing levels of confusion with increasing wavelength,
both because of the increase in the density of galaxies of a given
flux density and because of the increase in the diffraction limited
beam size. At 60
m, integrations greater than 10,000 seconds are
required to approach serious levels of confusion, while 200 seconds
and 100 seconds are required respectively at 100 and 150m. It is
possible with an 85 cm telescope to reach limits about two orders
of magnitude deeper than the data in IRAS bands 3 and 4.
We note that the scaling of the noise sources, for a given and
roughly optimized method of source extraction, is 
for
point source confusion noise and photon noise and 
for
cirrus noise. Thus, the relative importance of these differing
noise sources will change only slowly with telescope aperture. Any
telescope designed to operate at fundamental detection limits in
the far infrared should be cold enough so that the photon noise
does not increase significantly compared with the natural
background.