RESUMO Esta pesquisa foi desenvolvida para avaliar o desempenho vegetativo de progênies de cupuaçuzeiro, na fase de imaturidade, por meio da análise de medidas repetidas no tempo. A área experimental foi instalada no município de Tomé-Açu, Nordeste do Estado do Pará. Foram empregadas 25 progênies de irmãos completos de cupuaçuzeiro, que tiveram o desenvolvimento vegetativo (altura e diâmetro da planta) monitorado durante três anos. Por conta da natureza longitudinal das observações, foi avaliado primeiramente, por meio do teste de esfericidade de Mauchly, qual o tipo de análise estatística deveria ser aplicada. Para a variável diâmetro, a condição de esfericidade não foi violada, sendo assim, procedeu-se à análise no esquema de delineamento de parcelas subdivididas no tempo. Para a variável altura, o teste foi significativo, indicando necessariamente avaliar qual a melhor estrutura que explique a correlação dos erros, sendo escolhida a estrutura de covariâncias Simetria Composta Heterogênea. Foram encontradas diferenças entre as progênies somente no terceiro ano de avaliação, sendo que a variável diâmetro da planta permitiu discriminar melhor as progênies avaliadas.

ABSTRACT This study was conducted to evaluate vegetative performance of cupuassu progenies, in the immaturity phase, through analysis of repeated measures. The experimental area was installed in the municipality of Tomé Açu, Northeast of the state of Para. We employed 25 full-sib progeny of cupuassu, who had vegetative growth (height and diameter of the plant), monitored for three years. Because of the nature of longitudinal observations it was assessed primarily through Mauchly sphericity test, which type of statistical analysis should be applied. For the variable diameter, sphericity conditionwas not violated, therefore, proceeded to review the design scheme of split plot. For the height variable, the test was significant, indicating necessarily evaluate the best structure to explain the correlation of errors, being chosen the covariance structure Heterogeneous Composed Symmetry. Differences were found between the progenies only in the third year of evaluation, and the variable diameter of the plant allowed better discriminate the progenies.

The present work uses multivariate statistical analysis as a form of establishing the main sources of error in the Quantitative Phase Analysis (QPA) using the Rietveld method. The quantitative determination of crystalline phases using x ray powder diffraction is a complex measurement process whose results are influenced by several factors. Ternary mixtures of Al2O3, MgO and NiO were prepared under controlled conditions and the diffractions were obtained using the Bragg-Brentano geometric arrangement. It was possible to establish four sources of critical variations: the experimental absorption and the scale factor of NiO, which is the phase with the greatest linear absorption coefficient of the ternary mixture; the instrumental characteristics represented by mechanical errors of the goniometer and sample displacement; the other two phases (Al2O3 and MgO); and the temperature and relative humidity of the air in the laboratory. The error sources excessively impair the QPA with the Rietveld method. Therefore it becomes necessary to control them during the measurement procedure.