Quantitative Fiber Mixture Analysis by Scanning Electron Microscopy: Part VII: Modeling the Microscopic Analysis of Binary Animal Fiber Blends

摘要

During the Nice 2000 meeting of the Technology and Standards Committee of the International Wool Textile Organization, the test method "Scanning Electron Microscopic Analysis of Speciality Fibres and Sheep's Wool and their Blends" was installed as full test method TM 58. The 95%-confidence range of the blend composition q was identified as a suitable measure of accuracy. Due to practical limitations, only limited data were and will be available on the variability of q when retesting samples. To gain detailed insight into this variability beyond the limits of foreseeable availability of experimental data, namely through round trials, a computer model has been developed to reflect the practice of blend analysis according to TM-58. Modeling practically relevant blends with somewhat extreme fiber components confirms that the accuracy of the blend analysis, calculated on the basis of the Wildman/Bray formula, is in fact very robust. The modeled coefficients of variation of q do not exceed 10% and are generally smaller than those found, for example, in the 1995 CCMI round trial. Allowing, within reasonable limits, a somewhat higher variability of the confidence range in practice, we can conclude that the results corroborate the statements of TM-58 with respect to the accuracy of the methodology.