An exploratory project was conducted to assess the practical feasibility of establishing objective, empirically verified task descriptions or specifications (TS). Also, using the developed TS as a foundation, the possibility of erecting data structures (models) on them were investigated as were mathematical techniques for extracting information from these models. In Part I the concept of quasi-algorithms is introduced and applied in task analysis. An account is given of the development and successful verification procedures for quasi-algorithmic TS of a number of Field Artillery Fire Direction Center activities. These TS are listed. Improved procedures for routine development are outlined. Diverse uses of quasi-algorithmic TS, as such, are discussed. In Part II a mathematical model for structuring TS is presented. Abstract representation of TS in terms of attribute vectors is explained. A binary vector model is developed with reference to a TS from Part I. The concepts of dominance and cardinality are introduced, and it is shown how they structure any TS or set for TS. Applications to training, task taxonomy, duty assignments, career fields, etc. are outlined. The improved sensitivity of a model employing fuzzy subset theory is explained briefly.