Preliminary to making a series of timber tests it is very important that a working plan be prepared as a guide to the investigation. This should embrace: (1) the purpose of the tests; (2) kind, size, condition, and amount of material needed; (3) full description of the system of marking the pieces; (4) details of any special apparatus and methods employed; (5) proposed method of analyzing the data obtained and the nature of the final report. Great care should be taken in the preparation of this plan in order that all problems arising may be anticipated so far as possible and delays and unnecessary work avoided. A comprehensive study of previous investigations along the same or related lines should prove very helpful in outlining the work and preparing the report. (For sample working plan see Appendix.) FORMS OF MATERIAL TESTED In general, four forms of material are tested, namely: (1) large timbers, such as bridge stringers, car sills, large beams, and other pieces five feet or more in length, of actual sizes and grades in common use; (2) built-up structural forms and fastenings, such as built-up beams, trusses, and various kind of joints; (3) small clear pieces, such as are used in compression, shear, cleavage, and small cross-breaking tests; (4) manufactured articles, such as axles, spokes, shafts, wagon-tongues, cross-arms, insulator pins, barrels, and packing boxes. As the moisture content is of fundamental importance (see WATER CONTENT, above.), all standard tests are usually made in the green condition. Another series is also usually run in an air-dry condition of about 12 per cent moisture. In all cases the moisture is very carefully determined and stated with the results in the tables. SIZE OF TEST SPECIMENS The size of the test specimen must be governed largely by the purpose for which the test is made. If the effect of a single factor, such as moisture, is the object of experiment, it is necessary to use small pieces of wood in order to eliminate so far as possible all disturbing factors. If the specimens are too large, it is impossible to secure enough perfect pieces from one tree to form a series for various tests. Moreover, the drying process with large timbers is very difficult and irregular, and requires a long period of time, besides causing checks and internal stresses which may obscure the results obtained. On the other hand, the smaller the dimensions of the test specimen the greater becomes the relative effect of the inherent factors affecting the mechanical properties. For example, the effect of a knot of given size is more serious in a small stick than in a large one. Moreover, the smaller the specimen the fewer growth rings it contains, hence there is greater opportunity for variation due to irregularities of grain.