The traditional cosmological tests appear to have been overshadowed by observations of the anisotropies in the cosmic microwave background (CMB). We are told that these observations accurately measure the geometry of the Universe, its composition, its present expansion rate, and the nature and form of the primordial fluctuations . The resulting values for these basic parameters are very similar to those deduced earlier from a variety of observations - the so-called "concordance model" - with about 30% of the closure density of the Universe comprised of matter (mostly a pressureless, non-baryonic dark matter), the remainder being in negative pressure dark energy . Given the certainty and precision of these assertions, any current discussion of observational cosmology must begin with the question: Is there any room for doubt? Why should we bother with lower precision cosmological tests when we know all of the answers anyway?
While the interpretation of the CMB anisotropies has emerged as the single most important cosmological tool, we must bear in mind that the conclusions drawn do rest upon a number of assumptions, and the results are not altogether as robust as we are, at times, led to believe. One such assumption, for example, is that of adiabatic initial fluctuations - that is, 100% adiabatic. A small admixture of correlated isocurvature fluctuations, an aspect of braneworld scenarios , can affect peak amplitudes and thus, the derived cosmological parameters. A more fundamental assumption is that of the validity of traditional Friedmann-Robertson-Walker (FRW) cosmology in the post-decoupling universe. Is the expansion of the universe described by the Friedmann equation? Even minimal changes to the right-hand-side, such as the equation of state of the dark energy component, can alter the angular size distance to the last scattering surface at z = 1000 and the luminosity distance to distant supernovae. But even more drastic changes to the Friedmann equation, resulting from modified gravitational physics, have been proposed in attempts to remove the unattractive dark energy [4, 5].
Such suggestions reflect a general unease with the concordance model - a model that presents us with a universe that is strange in its composition. The most abundant form of matter consists of, as yet, undetected non-baryonic particles originally postulated to solve the problems of structure formation and of the missing mass in bound gravitational systems such as galaxies and clusters of galaxies. In this second respect, it is fair to say that it has failed - or, to be generous, not yet succeeded - because the predicted density distribution of dark halos which emerge from cosmic N-body  simulations appears to be inconsistent with observations of spiral galaxies  or with strong lensing in clusters of galaxies .
Even more mysterious is the "dark energy", the pervasive homogeneous fluid with a negative pressure which may be identified with the cosmological constant, the zero-point energy density of the vacuum. The problem of this unnaturally low energy density, 10-122 in Planck units, is well-known, as is the cosmic coincidence problem: why are we observing the Universe at a time when the cosmological constant has, fairly recently, become dynamically important ? To put it another way, why are the energy densities of matter and dark energy so comparable at the present epoch? This is strange because the density of matter dilutes with the expanding volume of the Universe while the vacuum energy density does not. It is this problem which has led to the proposal of dynamic dark energy, quintessence - a dark energy, possibly associated with a light scalar field - with an energy density that evolves with cosmic time possibly tracking the matter energy density . Here the difficulty is that the field would generally be expected to have additional observational consequences - such as violations of the equivalence principle at some level, possibly detectable in fifth force experiments .
For these reasons, it is even more important to pursue cosmological tests that are independent of the CMB, because one might expect new physics to appear as observations inconsistent with the concordance model. In this sense, discord is more interesting than concord; to take a Hegelian point of view - ideas progress through dialectic, not through concordance. It is with this in mind that I will review observational cosmology with emphasis upon CMB-independent tests.
Below I argue that the evolution of the early, pre-recombination universe is well-understood and tightly constrained by considerations of primordial nucleosynthesis. If one wishes to modify general relativity to give deviations from Friedmann expansion, then such modifications are strongly constrained at early times, at energies on the order of 1 MeV. However, cosmological evolution is much less constrained in the post-recombination universe where there is room for deviation from standard Friedmann cosmology and where the more classical tests are relevant. I will discuss three of these classical tests: the angular size distance test where I am obliged to refer to its powerful modern application with respect to the CMB anisotropies; the luminosity distance test and its application to observations of distant supernovae; and the incremental volume test as revealed by faint galaxy number counts.
These classical tests yield results that are consistent, to lower precision, with the parameters deduced from the CMB. While one can make minimal changes to standard cosmology, to the equation of state of the dark energy for example, which yield different cosmological parameters, there is no compelling observational reason to do so. It remains the peculiar composition and the extraordinary coincidences embodied by the concordance model that call for deeper insight. Such motivations for questioning a paradigm are not unprecedented; similar worries led to the inflationary scenario which, unquestionably, has had the dominant impact on cosmological thought in the past 25 years and which has found phenomenological support in the recent CMB observations.
I am not going to discuss cosmological tests based upon specific models for structure formation, such as the form of the luminous matter power spectrum  or the amplitude of the present mass fluctuations . I do not mean to imply that such such tests are unimportant, it is only that I restrict myself here to more global and model-independent tests. If one is considering a possibility as drastic as a modification of Friedmann expansion due, possibly, to new gravitational physics, then it is tests of the global curvature and expansion history of the Universe that are primary.
I am also going to refrain, in so far as possible, from discussion of theory - of new gravitational physics or of any other sort. The theoretical issues presented by dark matter that can only be detected gravitationally or by an absurdly small but non-zero cosmological constant are essentially not problems for the interpretive astronomer. The primary task is to realistically access the reliability of conclusions drawn from the observations, and that is what I intend to do.