Several intermediate reports on the progress in this project have already been published (Suehrcke, Harris & McBain 1996; McBain 1997a, 1998; McBain, Close, Suehrcke, Harris & Brandemuehl 1998; McBain & Harris 1998; McBain, Harris, Close & Suehrcke 1998; Suehrcke & McBain 1998). It is a pleasure to acknowledge the assistance of my coauthors in preparing these papers; their direct contributions are noted where utilized.
I would also like to express my gratitude to Prof. Don Close for pointing out the relevance of the physics of rainclouds to the question of interior condensation (§2.1.1); Prof. Bob Street for pointing out the connection between Theorem 1 and Hopf's Maximum Principle; Dr Jonathan Harris for informing me of the work of Daniels (1985; Daniels & Wang 1994) on the destruction of the conduction regime (ch. 5); Dr Wolfgang Schöpf for indicating the importance of hydrodynamic stability and providing several of the relevant references, as discussed in §4.6.2; Dr Harris for making available accurate finite element solutions of the creeping buoyant flow in a cuboid which proved very illuminating in the development of the material on the narrow cavity limit for bounded sections (ch. 7); Dr Harry Suehrcke for thorough checking of my derivations; and Dr Eric Peterson for interesting discussions on the inertial generation of axial flow by circulation in aquaculture tanks and provision of some of the references cited in §8.1.1. Thanks also to Roger Hosking, John Patterson, Lance Bode, Patrick Le Quéré and Andrew Schultz for enlightening conversations on applied mathematics and fluid mechanics.
I would also like to thank Drs Nick Stokes and Zili Zhu for prompt and thorough technical support with the Fastflo finite element package, used in chapter 5; Dr Harris and Mr Darrin Stephens for general and generous assistance with computers; and Drs Harris and Schöpf, and Mr Schultz for help with LATEX.
I am indebted to my supervisors (Prof. Close and Drs Suehrcke and Harris) and Ms Arian McVeigh for proofreading and helpful suggestions. I need hardly add that the remaining faults are entirely my own.
Finally I would like to thank Ms Kathleen Sharp for uncommonly painless administration of the financial assistance which enabled me to write this thesis.