The dark energy in the Universe is described in the context of modified Friedmann equations as a fluid parameterized by the density of dark matter and undergoing an adiabatic expansion. This formulation is applied to the Cardassian model. Choosing then parameters consistent with the supernova observations, it gives a background expansion in which the cosmic temperature fluctuations are calculated. The resulting spectrum is quite similar to what is obtained in the standard concordance model. If the Cardassian fluid is interpreted as a new kind of interacting dark matter, its overdensities are driven into oscillations when the interaction energy is rising in importance. This does not occur in a description of the Cardassian fluctuations motivated by theories of modified gravity. There the energy of the underlying matter is also conserved, which requires appearance of effective shear stress in the late Universe. In both approaches that allow fluctuations the thermal power spectrum at large scales is much too strongly enhanced by the late integrated Sachs-Wolfe effect. With the interacting dark matter assumption, we conclude that the Cardassian model is ruled out by observations, expect in a small neighbourhood of the $\Lambda$CDM limit.

Using the language of differential forms, the Kaluza-Klein theory in 4+1 dimensions is derived. This theory unifies electromagnetic and gravitational interactions in four dimensions when the extra space dimension is compactified. Without any ad-hoc assumptions about the five-dimensional metric, the theory also contains a scalar field coupled to the other fields. By a conformal transformation the theory is transformed from the Jordan frame to the Einstein frame where the physical content is more manifest. Including a cosmological constant in the five-dimensional formulation, it is seen to result in an exponential potential for the scalar field in four dimensions. A similar potential is also found from the Casimir energy in the compact dimension. The resulting scalar field dynamics mimics realistic models recently proposed for cosmological quintessence.

We show how the presence of a very light scalar with a cubic self-interaction in six dimensions can stabilize the extra dimensions at radii which are naturally exponentially large, $r \sim \ell \exp [(4\pi)^3/g^2]$, where $\ell$ is a microscopic physics scale and $g$ is the (dimensionless) cubic coupling constant. The resulting radion mode of the metric becomes a very light degree of freedom whose mass, $m \sim 1/(M_p r^2)$ is stable under radiative corrections. For $1/r \sim 10^{-3}$ eV the radion is extremely light, $m \sim 10^{-33}$ eV. Its couplings cause important deviations from General Relativity in the very early universe, but naturally evolve to phenomenologically acceptable values at present.

We examine the late-time (nucleosynthesis and later) cosmological implications of brane-world scenarios having large (millimeter sized) extra dimensions. In particular, recent proposals for understanding why the extra dimensions are so large in these models indicate that moduli like the radion appear (to four-dimensional observers) to be extremely light, with a mass of order 10^{-33} eV, allowing them to play the role of the light scalar of quintessence models. The radion-as-quintessence solves a long-standing problem since its small mass is technically natural, in that it is stable against radiative corrections. Its challenges are to explain why such a light particle has not been seen in precision tests of gravity, and why Newton's constant has not appreciably evolved since nucleosynthesis. We find the couplings suggested by stabilization models can provide explanations for both of these questions. We identify the features which must be required of any earlier epochs of cosmology in order for these explanations to hold.