This study revisits cassava cortex–sorbitol materials using six original numerical Raman exports and archived thermal records, with particular attention to spectral reliability and the distinction between thermal mass-loss and calorimetric signals. Untreated cassava cortex (CT100), formulations with 5, 15 and 25 wt.% sorbitol (CT5, CT15 and CT25), and sorbitol alone (SB100) were compared; chemically treated cortex (CTT100) was analysed separately. Each Raman export contained 2,090 intensity measurements between 100 and 4,278 cm⁻¹. A reproducible Savitzky–Golay and prominence-based workflow identified the strongest numerical maxima at 2,550, 2,246, 1,366, 2,444 and 1,218 cm⁻¹ for CT100, CT5, CT15, CT25 and SB100, respectively. These positions were confirmed directly from the supplied exports but are not necessarily assignable molecular bands; several high-wavenumber spikes require caution. Historical thermal graphs report features at 440, 400, 465, 410 and 370 °C, respectively. Their numerical correspondence with previously reconstructed derivative traces remains unverified; the historical temperatures are therefore presented as source-recorded observations, not independently validated DTG maxima. The Raman records demonstrate sample-dependent spectral variation, whereas the thermal evidence suggests a non-monotonic ordering of graph-reported features. The novelty lies in a transparent, reproducible reassessment of the original Raman measurements and an explicit evidence audit of their proposed thermal integration, extending an earlier thermal-focused report without implying new experiments or unverified structure–property relationships.