“…Our 1-D scalings depends mainly on whether d = zz or d = zx but also on the thermal evolution parameters ( , , ): reasonable values for ( , , ) are 1∕5 < < 1∕3, 1∕3 < < 2∕3, and 0 < < 1∕6 and are in agreement with all 3-D an 2-D numerical simulations assuming reasonable mantle rock rheologies and theoretical constraints from evolution models for the Earth [e.g., Deschamps and Sotin, 2000, and references therein;Deschamps et al, 2010;Lenardic et al, 2006;Moore, 2008;Parmentier and Sotin, 2000;Ratcliff et al, 1996;Sotin and Labrosse, 1999;Stamenković and Breuer, 2014;Wolstencroft et al, 2009;Yanagisawa and Yamagishi, 2005;Zhong, 2005]. Values of much smaller than 1/5 or even negative [Korenaga, 2003] have not been confirmed by 2-D or 3-D numerical experiments to date, and as such, we did not explore them.…”