Examinating the behaviors in the data dispersion, it is possible to recognize the Normal Distribution as a mathematic/statistic object of relevance to deepen its scholastic processing. This article proposes, from a historical-epistemological approach, to analyze and suggest an epistemic route for a conceptual approximation, with the aim of encouraging and multidisciplinary approach. Related to this, it is recognized the normal probability plot as an approximation to the Normal Distribution, centering the analysis in the behavior of the error's data beyond the data, getting a conceptual approximation external, singular and innovating from the one promoted in terms of curriculum. For that it is suggested to contextualize as a conceptual framework a traditional approach of the mathematic modeling from the mathematic didactic, clearing the use of the oxygen blood's saturation. Using the didactic engineering as qualitative methodological design, the answers to the modeling task applied in students from 17 to 18 years in the Libertador Bernardo O'Higgins region are analyzed. The results show that the modeling task is conductive to the learning of Normal Distribution, being today a proposal that increases didactically not only the teaching, but also the learning of the Normal Distribution.