Representative sampling forms a critical success factor for PAT. Instrumental analytical methods (X) are often based on "analysis by proxy", for which the representativity of reference data (Y) is critical for the underlying calibrations. Although the Theory of Sampling (TOS) has been known for more 50 years, it is still very little appreciated and implemented in today's analytical chemistry and Process Analytical Technologies even though sampling errors form the dominant part of what is (wrongly) termed "measurement errors" - and typically attain magnitudes 50- 100 X the analytical error s.s. There is an imperative need for understanding representative sensor - and reference sampling. Non-representative sampling can not be remedied by neither statistics, chemometric data analysis or by any other type of a posteori "correction"(sic). This lecture presents a theoretical model as well as practical framework of seven sampling unit operations (SUO's) with which to approach all types of sampling issues associated with PAT.