The dataset is intended to support the evaluation of soil texture estimation under field conditions by predicting soil textural fractions through multivariate linear modeling of textural reference parameters. It includes both measured soil texture data—percentages of clay, silt, and sand obtained through laboratory analysis—and estimated texture classes derived from the hand-feel modeling approach. To assess the accuracy of the field estimates, the Texture Estimation (TE) Score, developed by Franzmeier and Owens (2008), is applied. For modeling purposes, and due to the compositional nature of soil texture data (where clay, silt, and sand percentages sum to 100%), the dataset also includes additive log-ratio (ALR) transformed variables: ALR1, defined as log(silt/sand), and ALR2, defined as log(clay/sand). These transformations help preserve the statistical integrity of the compositional data analysis. This dataset serves as the supplementary material for the study "Fifty Years of Field Soil Texture Estimation in Italy: Accuracy, Challenges, and Improvements."

A spreadsheet containing 105 records of laboratory-based soil texture analyses with replicates, along with corresponding estimates of sand, silt, and clay fractions derived using the hand-feel soil texture approach. The estimation method is based on the textural reference parameters proposed by Judith K. Turk and Rebecca A. Young (2021), with the following key modifications: Use of wet sieving with a nylon mesh to improve field estimation of sand content. Standardization of the initial soil ball diameter to 3.5 cm to ensure consistency in tactile analysis and ribbon formation. Adjustment of certain tactile reference thresholds to enhance field usability. The textural reference parameters used include: Ribbon Length × Ribbon Strength (RL × RS); Wire Rating (WR); Grittiness Rating (GR), Grid Count (GC)