In the past two decades, cosmology has taken a promising course. Due to improved and new observational instruments and the observations made with them, a wealth of data has made possible the determination of cosmological parameters with higher precision than ever before ("precision cosmology"). On the theoretical side, the interaction of elementary particle physicists and astrophysicists has provided major contributions to the interpretation of observations. In spite of the progress made, the standard cosmological model, developed into the "concordance model", seems not to be in good shape. With 95% of the matter content of the universe presently being of an unknown nature, can any claim be made that today's cosmological model leads to a better understanding of the universe than the model of two decades ago?
In this situations it may not be contraproductive to inquire about the nature of the discipline. Here, we encounter a common endeavour of mathematics, theoretical physics, astronomy, astro-, nuclear and elementary particle physics with the aim of explaining more than the cosmogonic myths of our forefathers. Has cosmology become a natural science, even a branch of the exact sciences? It certainly is a field of research well established by all social criteria if we follow J. Ziman  and define natural science as an empirical science steered by public agreement among scientists. In this context, "empirical" means that conclusions are not merely drawn by rational thinking as in the humanities but that they are tested by help of reproducible quantitative experiments/observations. Data from these measurements are interpreted by consistent physical theories and receive a preliminary validation to be reconsidered in the light of new facts. Cosmology as a very young scientific discipline has not yet reached the same degree of differentiation as other subfields of physics. 1 Most of what follows will refer to physical cosmology on a solid empirical basis and to its subfield named here originative cosmology. In the latter, the speculative parts, necessarily implied by physical theorizing, are dominant; they are just beginning to be linked to empirical testing or still await probing in the future. 2
Three periods of extremely unequal duration in the time evolution of the expanding universe will be used for gaining an impression of cosmology. They are: The flashlike "very early universe" of Δ t ~ 10-12 s duration (before the assumed electroweak phase transition); it includes the inflationary era and prior Planck-scale modeling (quantum cosmology). Next, the "early universe" (until early structure formation) amounting to ~ 4% of the total age of the universe [(13.27 ± 0.12) ⋅ 109 y] and covering Δt ~ 4.3 ⋅ 108 years; here, nucleosynthesis and the release of cosmic background radiation (CMB) can be found. Finally, the remaining period from structure formation (reionization) until today comprising ~ 96% of the time. Einstein's theory of gravitation will be the almost exclusive theoretical background adopted here because its implications for physical cosmology have been developed best. In the following, I shall use the words "cosmos" and "universe" as synonyms although they carry different rings; cosmos goes well with order and coherence, while universe implies uniqueness and entirety. Before going into details of cosmological modeling I will try to circumscribe cosmology as a field of research.
1 This is reflected by the PACS-classification which provides only 7 subclasses for cosmology, 20 for solar physics, and 178 for "solid earth physics". Back.
2 An endeavour purporting to belong to physics but without any connection to an empirical background will be called make-believe cosmology, cf. section 5.3. This is to function as a reminder that the universe exists not just "on paper" as the philosopher P. Valéry would have it. Back.