Cognitive load theory seeks to generate novel instructional designs through a focus on human cognitive architecture including a limited working memory; however, the potential for enhancing learning through non-visual or non-auditory working memory channels is yet to be evaluated. This exploratory experiment tested whether explicit instructions to trace out elements of geometry worked examples with the index finger would enhance learning, as measured by error rates and problems solved. Compared to a non-tracing condition, students in the tracing condition correctly solved more practice problems, and made fewer errors on a subsequent test. Recommendations for subsequent more sensitive experiments are made.