MIL-HDBK-17-1F Volume 1,Chapter 2 Guidelines for Property Testing of Composites specimen and test procedure.is just as important as theoretical perfection.Robustness,or lack thereof. is assessed by interlaboratory testing,and is measured by precision(variation in the sample population) and bias(variation of the sample mean from the true average).'The precision and bias of test methods are evaluated by comparison testing (often called "round-robin"testing)both within-laboratory and be- tween laboratories.The obvious ideal is high precision(low variation)and low bias(sample mean close to true average)both within-laboratory and between laboratories.Such a test method would repeatedly produce reproducible results without regard to material,operator,or test laboratory.However,quantifica- tion of bias requires a material standard for each test;none of which are currently available for compos- ites.As a result,bias of composite test methods can currently only be qualitatively assessed. Somewhat separate from the precision and bias of a test method (for a given specimen)is the effect on precision and bias of variation in test specimen size and geometry.For heterogeneous materials, physically larger specimens can be expected to contain within the coupon a more representative sample of the material microstructure.While desirable,a larger specimen is more apt to contain a greater num- ber of micro-or macro-structural defects than a smaller specimen,and thus can be expected to produce somewhat lower strengths (though possibly also with lower variation).Variations in specimen geometry can also create differing results.Size and geometry effects can produce statistical differences in results independent of the "degree of perfection"of the remaining aspects of a test method or its conduct;such effects should be expected.Therefore,even though the specimen response may not (and probably won't)be identical to that of the structure,the "ideal"test method will incorporate a specimen geometry that can be consistently correlated with structural response As the criticality of various test parameters are still being researched and understood(even for rela- tively common tests)and as "standard laboratory practices,"upon close examination,are actually found to vary from laboratory to laboratory,it is critical to control or document as many of these practices and parameters as possible.ASTM Committee D-30,responsible for standardization of advanced composite material test methods,tries to consider all of these factors when improving existing and developing new standard test methods (see Reference 2.2.4).Due to both their completeness and their status as full- consensus standards,ASTM D-30 test methods,where applicable,are emphasized by this handbook. Failure to minimize test method sensitivities,whatever the cause,can cause the statistical methods contained within MIL-HDBK-17 to break-down,as all variation in data is implicitly assumed by the statisti- cal methods to be due to material or process variation.Any additional variation due to specimen prepara- tion or testing procedure is added to the material/process variation,which can result in extraordinarily conservative.or even meaningless.basis value results. Test methods,with emphasis on ASTM standards for advanced composites,are discussed in Chap- ters 3 through 7.The advantages and disadvantages of the various test methods for composites are dis- cussed,including.for completeness,non-standard but often referenced methods that have appeared in the literature.Chapters 3 and 4 cover constituent testing.Chapter 5 covers prepreg test methods. Chapter 6 covers lamina and laminate testing.Chapter 7 covers structural element test methods.Data produced by the following test methods (Table 2.2.4)are currently being accepted by MIL-HDBK-17 for consideration for inclusion in Volume 2. The term"accuracy"is often used as a generic combination of aspects of both precision and bias.The terms "precision"and "bias",being more specific,are preferred for use where appropriate 2-8MIL-HDBK-17-1F Volume 1, Chapter 2 Guidelines for Property Testing of Composites 2-8 specimen and test procedure, is just as important as theoretical perfection. Robustness, or lack thereof, is assessed by interlaboratory testing, and is measured by precision (variation in the sample population) and bias (variation of the sample mean from the true average).1 The precision and bias of test methods are evaluated by comparison testing (often called "round-robin" testing) both within-laboratory and be￾tween laboratories. The obvious ideal is high precision (low variation) and low bias (sample mean close to true average) both within-laboratory and between laboratories. Such a test method would repeatedly produce reproducible results without regard to material, operator, or test laboratory. However, quantifica￾tion of bias requires a material standard for each test; none of which are currently available for compos￾ites. As a result, bias of composite test methods can currently only be qualitatively assessed. Somewhat separate from the precision and bias of a test method (for a given specimen) is the effect on precision and bias of variation in test specimen size and geometry. For heterogeneous materials, physically larger specimens can be expected to contain within the coupon a more representative sample of the material microstructure. While desirable, a larger specimen is more apt to contain a greater num￾ber of micro- or macro-structural defects than a smaller specimen, and thus can be expected to produce somewhat lower strengths (though possibly also with lower variation). Variations in specimen geometry can also create differing results. Size and geometry effects can produce statistical differences in results independent of the "degree of perfection" of the remaining aspects of a test method or its conduct; such effects should be expected. Therefore, even though the specimen response may not (and probably won't) be identical to that of the structure, the "ideal" test method will incorporate a specimen geometry that can be consistently correlated with structural response. As the criticality of various test parameters are still being researched and understood (even for rela￾tively common tests) and as "standard laboratory practices," upon close examination, are actually found to vary from laboratory to laboratory, it is critical to control or document as many of these practices and parameters as possible. ASTM Committee D-30, responsible for standardization of advanced composite material test methods, tries to consider all of these factors when improving existing and developing new standard test methods (see Reference 2.2.4). Due to both their completeness and their status as full￾consensus standards, ASTM D-30 test methods, where applicable, are emphasized by this handbook. Failure to minimize test method sensitivities, whatever the cause, can cause the statistical methods contained within MIL-HDBK-17 to break-down, as all variation in data is implicitly assumed by the statisti￾cal methods to be due to material or process variation. Any additional variation due to specimen prepara￾tion or testing procedure is added to the material/process variation, which can result in extraordinarily conservative, or even meaningless, basis value results. Test methods, with emphasis on ASTM standards for advanced composites, are discussed in Chap￾ters 3 through 7. The advantages and disadvantages of the various test methods for composites are dis￾cussed, including, for completeness, non-standard but often referenced methods that have appeared in the literature. Chapters 3 and 4 cover constituent testing. Chapter 5 covers prepreg test methods. Chapter 6 covers lamina and laminate testing. Chapter 7 covers structural element test methods. Data produced by the following test methods (Table 2.2.4) are currently being accepted by MIL-HDBK-17 for consideration for inclusion in Volume 2. 1 The term "accuracy" is often used as a generic combination of aspects of both precision and bias. The terms "precision" and "bias", being more specific, are preferred for use where appropriate