This paper reconsiders Persson and Tabellini's (2003, 2004) analysis of the causal effect of constitution type on government size. It addresses the concerns of Acemoglu (2005) and makes some measurement and methodological refinements to the identification strategy to argue there is a qualitatively large and statistically significant relationship between constitution type and government size. The age of a democracy is of increased importance in the new identification strategy, but existing measures of when countries became democracies are shown to be flawed. Two new measures of the age of a democracy are introduced. The first details when a country first had a genuinely democratic election, the second when its current constitution was promulgated.