SUMMARY
The objective was to conduct a collaborative study to evaluate the mineral matter contents in feeds in seven feed analysis laboratories from institutions linked to the National Institute of Science and Technology in Animal Science (INCT-CA). Six feeds were evaluated: signal grass hey, sugarcane, corn silage, soybean meal, corn grain and citrus pulp. It was asked to each laboratory that the samples were analyzed for six days, consecutive or not, with three replicates per sample per day of evaluation. Variations were observed among laboratories, days of analysis and interaction effect of material and laboratory. The variation of results among laboratories accounted from 33.5 to 78.2% of the total random variation. It was verified that the repeatability represented 8.5 to 51.6% of reproducibility. It was noticed Horwitz ratios greater than 2 for three materials tested. It was found that 42.9% of the assessments show variations of the estimates as a function of days in which analysis was performed. Differences were also observed between procedures to assess mineral matter according to the different laboratories. The mineral matter contents estimate by each laboratory in this study have low reproducibility, which is aggravated by the interaction effect of material and laboratory.
Keywords:
analytical methods; ash; feed analysis
