Two kinds of work-hardening are discussed, interaction-hardening and source-hardening. Interaction-hardening arises because a large stress must be applied to move a dislocation through a material which already contains a large number of dislocations. This paper is mainly concerned with source-hardening which occurs because the sources of easy glide become inactive. Additional glide occurs by the use of sources which act at a higher stress level. Source-hardening is the most important kind of hardening at small strains since then the dislocation density is small.The experimental stress-strain curves obtained by Rosi and Mathewson on pure aluminum crystals can be calculated theoretically as follows: At very low temperatures one assumes that one has a distribution of "free lengths" of dislocations. The Frank-Read mechanism for the production of dislocations is used. It is also assumed in accordance with electron microscope data that generation ceases after about 500 dislocation loops have been produced. Source-hardening therefore occurs as the sources of greatest "free length" are gradually used. At finite temperatures the following modifications are introduced: The total stress is the sum of the applied stress and the stress resulting from thermal fluctuations. In addition, Brown's electron microscope data indicates that a given source produces much more glide at high temperature than at low temperature. The present theory does not predict an equation of state and does check Rosi and Mathewson's data on this topic.Suggestions are made to explain the reason for the increased glide per source with increased temperature and to understand the way in which a source becomes inactive as a result of vacancy production.Numerous experiments are suggested.The major accomplishment of the paper is that it is able to show that the lamella structure found in the electron microscope pictures can be related to the mechanical properties of the metal single crystals.