The purpose of atmospheric corrosion testing is two-fold: to determine (1) the rate and nature of corrosive attack of a metal by atmospheric agencies and (2) the efficacy of measures used for preventing the attack. Results of such tests should always be correlated with the prevailing atmospheric conditions at the test site. Conclusions supported by tests at one location do not always hold for other locations. The degree and nature of the pollution of the air is a factor which must be very carefully considered. The importance of this factor is often not sufficiently recognized. Specimens having a large area-mass ratio are recommended. This requirement limits the tests in large measure to sheet material. The specimens mounted on rigid permanent racks usually face the south at an angle of approximately 30 deg. to the horizontal. Except in special tests, no specimen should be influenced by another by contact, rain drip or otherwise. Contact of specimens with supports should be minimized by the use of grooved holders of porcelain or other inert material. Quantitative results are based on measurements of change of weight (usually loss of weight, per unit area per unit of time, after cleaning the specimen) or change of strength or other physical property. Loss of ductility is generally considered as being of most significance. Observations on the amount and distribution of surface corrosion can be expressed semiquantitatively. Size of specimens and frequency of inspection vary with the purpose of the test. As large a specimen as practicable is recommended in tests for obtaining engineering corrosion data in which visual inspection is depended upon in the inspection. For change of weight determination, sheet specimens as large as 9 by 12 in. can be used. For change of strength determination conventional test specimens can be used although for some materials it is advisable to machine the tension specimens from the sheet after corrosion. Although no attempt has been made to formulate a recommended practice to be followed in corrosion testing of this kind, a number of factors have been indicated that should be given consideration in any movement along this line.