This paper examined the number magnitude understanding and estimation of elementary students. Children as early as kindergartner differ in their understanding of numerical magnitudes. Research has found that teaching linearity of numbers was effective in improving children's estimation of numerical magnitudes (Booth & Siegler, 2008). Extending this line of work, I examined two different ways that linearity of numbers is taught: multiple 10 blocks and bundles. There were 58 kindergarten and first-grader students from Southern California randomly assigned to one of the three instructional contexts: blocks, bundles and control condition. I found that the students using the combination of base-10 and unit blocks gained more from the instruction than the other two groups. The students in the blocks group presented linearity for mental number line after the instruction. However, those in bundle and control groups could not.