A critical examination of Popper’s falsificationism as a methodological criterion of demarcation led to the development o f a supplementary means of distinguishing science from pseudo- science The discipline is made the unit of appraisal and its pattern o f historical development b used as the indicator of demarcation. Results of a test of this indicator against astrology and physical optics accord with our basic judgm ents of these disciplines. The indicator effectively reveals that scientific creationism is pseudo-science, and that evolutionary biology is genuine science. Three fundam ental approaches to scientific investigation, viz. v erificationism , falsificationism and m ulti-cornered testing (M CT) are contrasted. MCT is distinguished by competition between hypotheses, which makes it more informative than at least the naive versions of the other two approaches. While competition does not produce immediate victors, it does make demands on theories, which can be augmented by prescribing a series of independent tests. The comparative method implies the existence of two types of evidence. Common evidence is that which io predicted or explained by two or more rival hypotheses. Discriminatory evidence favours one rival over the others. It is argued that in both the fields of species biology and speciation there have been instances o f over-relying on common evidence, o f indistinctly defining alternative hypotheses, of ro t following their logical consequences and of not using exisiing discriminatory evidence to adjudicate between these hypotheses. Species concepts and definitions of modes o f speciation are evaluated. Normative principles are suggested for defining species and other important terms in evolutionary biology, and for testing species concepts and modes of speciation. The advantages and limitations o f a historical indicator of demarcation and the merits and principles of the comparative approach to method are discussed and illustrated using the analoev of a mathematical game. Scientific crcanomsni is shown to have a coating of scientific method, but to have systematically violated fundamental methodological principles. D arn in ’* method in contrast, had a comparative structure, and distinguished between common *nd discriminatory evidence. While there are methodological problems sn evolutionary biology, these are shown to be minor in comparison to that four*! in to c n o fk ciratxxiiun.

"Because many software systems used for business today are considered legacy systems, the need for software evolution techniques has never been greater. We propose a novel evolution methodology for legacy systems that integrates the concepts of features, regression testing, and Component-Based Software Engineering (CBSE). Regression test suites are untapped resources that contain important information about the features of a software system. By exercising each feature with its associated test cases using code profilers and similar tools, code can be located and refactored to create components. The unique combination of Feature Engineering and CBSE makes it possible for a legacy system to be modernized quickly and affordably. We develop a new framework to evolve legacy software that maps the features to software components refactored from their feature implementation. In this dissertation, we make the following contributions: First, a new methodology to evolve legacy code is developed that improves the maintainability of evolved legacy systems. Second, the technique describes a clear understanding between features and functionality, and relationships among features using our feature model. Third, the methodology provides guidelines to construct feature-based reusable components using our fine-grained component model. Fourth, we bridge the complexity gap by identifying feature-based test cases and developing feature-based reusable components. We show how to reuse existing tools to aid the evolution of legacy systems rather than re-writing special purpose tools for program slicing and requirement management. We have validated our approach on the evolution of a real-world legacy system. By applying this methodology, American Financial Systems, Inc. (AFS), has successfully restructured its enterprise legacy system and reduced the costs of future maintenance. "