It is generally accepted¹ that scientific knowledge is achieved when a theory (in fact, only its consequences) is directly confronted with observations (or experiments)². Since the time of Galileo Galilei (1564-1642), observation and/or experimentation has been used to confirm (continuously) or to falsify (forever) theories. No theory can be considered scientific without such crucial tests. To accept a theory without this continuous experimental/observational validation is to accept it as a dogma.
In a paper of Matravers, Ellis and Stoeger³ (p.31) they said: In fact, a rather serious and disturbing situation has developed within modern cosmology (I should add - in quantum field theory, too), in which some workers promote certain cosmological theories as correct and well-established without seeming to regard the adequacy of their observational or experimental justification as any importance. At the same time they tend to dismiss more observationally-based approaches - for example the kind of larger justificatory investigation we have just proposed - as being unnecessary or even `unscientific', simply because such approaches do not unquestioningly incorporate the standard view. This attitude is itself dangerously close to being unscientific, for it elevates theory above observation and relies on simplified geometrical models (certainly of considerable explanatory power⁴ without subjecting them to adequate observational testing - or even denying that they should be tested. Other authors like Tolman, MacCallum, Wesson, Rothman and Ellis, Krasiński, Vaucouleurs, Ribeiro and Videira⁵, have also expressed their sense of disagreement for the fictitious properties of simplified (and therefore nonexisting) universes and apparent loss of contact with empirical evidence and observational facts. To accept a theory without experimental or observational validation is to accept it as a dogma. However, things are not so simple.
At the end of the last century Boltzmann was engaged in a defense of the viewpoint that all scientific theories were nothing more than representations of the real phenomena. Whether his viewpoint is true or not cannot be proven inside the body of science, so it also cannot be refuted by science. While this viewpoint agrees with the freedom of personal choice in the theoretical work, it also denies the idea that we can ever achieve an ultimate knowledge of nature through science.
Thus, Boltzmann⁶ was engaged in a defense of the atomic concept (although his view-point of an atom was different than it is nowadays) which, at that time, was facing a growing number of opponents, like Ostwald⁷, who considered the atomic picture of the world to be outdated. He advocated its replacement by the view that the real world could be correctly described only by means of the concept of energy conservation and its derivatives which implied the denial of atomic idea⁸. Boltzmann feared that such a purely energetic representation would lead physics to become dogmatic, a fact that would also inevitably lead to its stagnation. This led him to two main theses⁹. In the first one, he stated that there was nothing more in a physical theory than a representation of nature to our minds. The second one stated consequently that nature could be represented by many different theories, which could even oppose each other. These theses were only epistemological, and could neither be proven or disproven within science which was empirical. Boltzmann emphasized that scientific work was impossible without using the theoretical concepts (basic objects and axioms), which stemmed from the fact that it was impossible to formulate any scientific theory simply from the mere observation of real phenomena.
At present, even the slightest mention touching the relation between science and faith can readily be perceived with distaste. However, it should be obvious that at the base of almost (this ''almost'' is due to the Only God alone who, as the only one, can prove assumptions) every field of science there lie (more or less numerous and more or less explicitly stated) assumptions, i.e. basic notions and axioms, which exist outside the possibility of proving them within the body of a given field of science. If we forget that these assumptions were taken for granted without any proof, and bring into science a language connected with these assumptions as the only one corresponding with the reality, then we can loose other characteristics of science - sincerity and freedom. I mean here especially the sincerity in declaring openly what is outside the proof. With sincerity and freedom suppressed, a given field of science automatically becomes a confession of faith, which is exactly what science tries to avoid. Consequences are always the same: a branch of science touched by such suppression of sincerity looses its freedom, because any other formulation, based on different assumptions, is automatically treated as a competitive confession of another faith. Such dishonest science very often confuses assumptions with their consequences, claiming that it proves what it actually is not able to prove as lying at its very base -- unprovable by definition.
At the same time honest investigation of consequences of basic assumptions for different formulations of any particular field of science should take place in the sphere of examining the conformity of these consequences with observations. This is the principle of objectiveness which should characterize science. I conclude that in the body of science the following pattern is allowed: science from faith (which goes before assumptions of the model), but not faith from science.

References:
1. In its first part, this text follows the article written by M.B. Ribeiro and A.A.P. Videira, "Dogmatism and Theoretical Pluralism in Modern Cosmology", preprint, arXir: physics/9806011, (1998).
2. K.R. Popper, "The Logic of Scientific Discovery", (1959). There are other publications on the topic.
3. D.R. Matravers, G.F.R. Ellis, W.R. Stoeger, Q. J. R. astr. Soc. 36:29, (1995).
4. I do not agree with the statement that these models have considerable explanatory power; see: J. Syska, "Self-consistent classical fields in gauge field theories", PhD thesis, (Institute of Physics, University of Silesia, Poland).
5. R.C. Tolman, Proc. Nat. Acad. Sci. (Wash.) 20:169, (1934);
   G. de Vaucouleurs, Science 167:1203, (1970);
   T. Rothman and G.F.R. Ellis, Astronomy 15, No.2:6, (1987);
   A. Krasiński, "Inhomogeneous Cosmological Models", (Cambridge University Press), part IX, Afterthoughts, (1997);
   M.B. Ribeiro and A.A.P. Videira, "Dogmatism and Theoretical Pluralism in Modern Cosmology", preprint, (1998).
6. L. Boltzmann, (1897), "On the Indispensability of Atomism in Natural Science", pp.41-53, and "Theoretical Physics and Philosophical Problems: Selected Writings", ed B. McGuinness, (Reidel, Dordretcht, 1974).
7. W. Ostwald, "Die Ueberwindung des wissenschaftlichen Materialismus", in "In Abhandlungen und Vortraege, Allgemeines Inhaltes (1887-1903)", pp.220-240, (Veit. and Co., Leipzig).
8. ibid., A.A.P. Videira, "Atomism and Energetics at the End of the 19th Century: the Luebeck Meeting of 1895", preprint CBPF-CS-003/95, (1995).
9. ibid., A.A.P. Videira, "Atomisme Epistemologique et Pluralisme Theorique dans la Pensee de Boltzmann", PhD thesis, Equipe Rehseis (CNRS), (University of Paris VII).