This study quantified partial seed coat loss from almond kernels. Basic physical properties of Nonpareil, Monterey, and Butte-Padre kernels were measured to determine which correlate to surface area, and could therefore predict it. Rehydrated almonds were manually peeled and images of seed coats were digitally analyzed. Surprisingly, individual dimensions (length, width, thickness) did not increase with increasing surface area, nor they did scale in proportion to one other. Almond surface area is often estimated from an equivalent sphere, but the sphere-based estimate only predicted 60% of the variation in measured surface area. An empirical model was created to predict surface area (r(2) = 0.74), based on the almond variety, as well as length, width, and mass after rehydration. By comparing the predicted total surface area and the measured surface area of any remaining seed coat, a quantitative percentage of lost seed coat can be calculated. (C) 2014 Elsevier Ltd. All rights reserved.