We present measurements of the auto- and cross-frequency power spectra of the cosmic infrared background (CIB) at 250, 350, and 500um (1200, 860, and 600 GHz) from observations totaling ~ 70 deg^2 made with the SPIRE instrument aboard the Herschel Space Observatory. We measure a fractional anisotropy dI / I = 14 +- 4%, detecting signatures arising from the clustering of dusty star-forming galaxies in both the linear (2-halo) and non-linear (1-halo) regimes; and that the transition from the 2- to 1-halo terms, below which power originates predominantly from multiple galaxies within dark matter halos, occurs at k_theta ~ 0.1 - 0.12 arcmin^-1 (l ~ 2160 - 2380), from 250 to 500um. New to this paper is clear evidence of a dependence of the Poisson and 1-halo power on the flux-cut level of masked sources --- suggesting that some fraction of the more luminous sources occupy more massive halos as satellites, or are possibly close pairs. We measure the cross-correlation power spectra between bands, finding that bands which are farthest apart are the least correlated, as well as hints of a reduction in the correlation between bands when resolved sources are more aggressively masked. In the second part of the paper we attempt to interpret the measurements in the framework of the halo model. With the aim of fitting simultaneously with one model the power spectra, number counts, and absolute CIB level in all bands, we find that this is achievable by invoking a luminosity-mass relationship, such that the luminosity-to-mass ratio peaks at a particular halo mass scale and declines towards lower and higher mass halos. Our best-fit model finds that the halo mass which is most efficient at hosting star formation in the redshift range of peak star-forming activity, z ~ 1-3, is log(M_peak/M_sun) ~ 12.1 +- 0.5, and that the minimum halo mass to host infrared galaxies is log(M_min/M_sun) ~ 10.1 +- 0.6.

We present spectroscopic and eleven-band photometric redshifts for galaxies in the 100-uJy Subaru/XMM-Newton Deep Field radio source sample. We find good agreement between our redshift distribution and that predicted by the SKA Simulated Skies project. We find no correlation between K-band magnitude and radio flux, but show that sources with 1.4-GHz flux densities below ~1mJy are fainter in the near-infrared than brighter radio sources at the same redshift, and we discuss the implications of this result for spectroscopically-incomplete samples where the K-z relation has been used to estimate redshifts. We use the infrared--radio correlation to separate our sample into radio-loud and radio-quiet objects and show that only radio-loud hosts have spectral energy distributions consistent with predominantly old stellar populations, although the fraction of objects displaying such properties is a decreasing function of radio luminosity. We calculate the 1.4-GHz radio luminosity function (RLF) in redshift bins to z=4 and find that the space density of radio sources increases with lookback time to z~2, with a more rapid increase for more powerful sources. We demonstrate that radio-loud and radio-quiet sources of the same radio luminosity evolve very differently. Radio-quiet sources display strong evolution to z~2 while radio-loud AGNs below the break in the radio luminosity function evolve more modestly and show hints of a decline in their space density at z>1, with this decline occurring later for lower-luminosity objects. If the radio luminosities of these sources are a function of their black hole spins then slowly-rotating black holes must have a plentiful fuel supply for longer, perhaps because they have yet to encounter the major merger that will spin them up and use the remaining gas in a major burst of star formation.