There is a mistake in footnote 5 on p.1231a, describing how to modify second-order direct inversion so that the corresponding balanced model conserves mass locally. The footnote should read as follows:

⁵First substitute (3.3c) into (3.3e) and into the first and last terms on the right of (3.3b), then delete div^-1delta from (3.3c). This makes (3.3) with DQ/Dt = 0 into the locally-mass-conserving version of the Bolin-Charney balance equations, as discussed, for example, by Gent and McWilliams (1984) and Whitaker (1993).