The first part of this thesis discusses the types and the properties of errors, one of which makes up systematic errors of measurements, removable by detecting their causes, the other errors of accidental causes which can not be removed. The final part of this thesis deals with the historical background of the Gaussian distribution by Hershel, Hagen, Laplace and Gauss from the late 18th century to the early 19th century. It can be concluded that the accidental idea and the treatment of accidental error distribution by Gauss Is the best one based on the assumption that the most probable value of true value is the arithmetic mean of data, obtained by repeated measurements of a given quantity.