Kant's claim that arithmetical truths are synthetic is famously contradicted by Frege, who considers them to be analytical. It may seem that this is a mere dispute about linguistic labels, since both Kant and Frege agree that arithmetical truths are a priori and informative, and, therefore, it is only a matter of how one chooses to call them. I argue that the choice between calling arithmetic “synthetic” or “analytic” has a deeper significance. I claim that the dispute is not a (...) merely linguistic one, but one involving the status of arithmetic as a science. Namely, it is a dispute about what truths are appropriate to use in grounding a scientific discipline. I interpret Frege's critique of Kant as being mainly about the lack of distinction between the context of justification and the context of discovery. (shrink)