Understanding a chemical phenomenon across the macroscopic, submicroscopic and symbolic domains (chemistry triplet) is the central tenet of chemistry thinking (Talanquer, 2018). This study utilised drawing as a potential tool to cultivate students’ understanding of the chemistry triplet (Johnstone, 1997) in learning chemical reactions. In drawing reactions, some are easier than others depending on whether they involved atoms and/or molecules only, or ions. However, it is not clear why difficulty in drawing reactions involving ions differs. To address this, my study investigated how teacher-feedback and peer-feedback, two common sources of feedback in classroom teaching, affected students’ drawings of different reactions at the submicroscopic level. A two-group quasi-experimental research design was used. Two Grade 11 (~17 years old) General Chemistry classes from a public laboratory high school in the Philippines received either teacher-feedback (50 students) or peer-feedback (51 students). The students in both groups performed a series of practical work on the four types of chemical reactions (i.e., combination, decomposition, single displacement, and double displacement), drew freehand their submicroscopic understanding of these reactions, received feedback and revised their drawings. The ICAP theory (Chi, 2021, 2009; Chi & Wylie, 2014) was applied to characterise student engagement with the feedback source. Based on the theory, teacher-feedback was regarded as active learning and peer-feedback as interactive learning. A month after the intervention, students’ learning outcomes in terms of recall and transfer were assessed. Students’ drawings were analysed using a visual-spatial scoring rubric. The findings showed that the difficulty of drawings reactions did not depend on the reaction types but on the identity of the chemical species involved, specifically the features of the ionic species. Also, quantitative results demonstrated that teacher-feedback supported students to produce higher quality of drawings than peer-feedback, particularly for the more challenging reactions, i.e., involving identity transformation (atoms and/or molecules becoming ions or vice versa). The delivery of teacher-feedback during whole-class presentation and peer-feedback during in-group discussion was audiotaped and videotaped. The feedback dialogues were analysed using an iterative cycle of deductive and inductive analysis. Qualitative results demonstrated that the two feedback sources exhibited distinct categories of feedback foci and feedback moves. These categories explicated feedback features beneficial for supporting drawing to learn for large classes. These features include adequate level of conceptual knowledge of the feedback provider, use of explanations and posing questions as productive feedback moves, and active student engagement with feedback. The findings of my study offer a suggestion on how to sequence progressively the teaching of different types of chemical reactions.