| The basic objective of this thesis is to prove interior and boundary continuity for solutions of degenerate elliptic partial differential equations of divergence type. To accomplish this a number of weighted inequalities are proven, including a characterization of weights for several Sobolev inequalities. A much larger class of degeneracies is considered than previously dealt with in the literature, and results are established which until now have been known only in the strongly elliptic case.;(VBAR)A(VBAR) (LESSTHEQ) (mu)(x)(VBAR)(DEL)u(VBAR)('p-1) + a(,1)(x)u('p-1) + a(,2)(x),;A (.) (DEL)u (GREATERTHEQ) (lamda)(x)(VBAR)(DEL)u(VBAR)('p) - c(,1)(x)u('p) - c(,2)(x),;(VBAR)B(VBAR) (LESSTHEQ) b(,0)(lamda)(x)(VBAR)(DEL)u(VBAR)('p) + b(,1)(x)(VBAR)(DEL)u(VBAR)('p-1) + b(,2)(x) u('p-1) + b(,3)(x).;Two approaches are taken to the problem of establishing continuity. The first uses a Harnack inequality and the second, Morrey spaces. The first applies to equations of the form div A = B, where A, B satisfy the growth conditions.;A Harnack inequality is proven for weights (mu), (lamda) satisfying certain capacitary conditions. Interior continuity follows immediately from this, and a Wiener criterion is established for continuity at the boundary. This generalizes work of Edmunds and Peletier, Gariepy and Ziemer, Murthy and Stampacchia, Trudinger, and Kruzkov.;A theory of weighted Morrey spaces is developed which establishes continuity estimates for a wide class of weighted Sobolev spaces W(,(omega))('1,p) with p>n, n the spatial dimension. This is in turn applied to prove the continuity of solutions of systems of the form div A(,i) = B(,i), i = 1 , ... , m, where A(,i) and B(,i) satisfy growth conditions similar to the above with p > n-(epsilon). In the nondegenerate case this is due to Widman and, in more general form, to N. Meyers and A. Elcrat.;The basic tools developed in this thesis to accomplish the above are: (1) A characterization of the weights (omega), (nu), (lamda), in terms of capacities for the inequalities ((INT) u('q) d(omega))('1/q) (LESSTHEQ) c((INT) (VBAR)(DEL)u(VBAR)('p) d(nu))('1/p) and ((INT) (VBAR)u - (INT) u d(lamda)(VBAR)('q) d(omega))('1/q) (LESSTHEQ) c((INT) (VBAR)(DEL)u(VBAR)('p) d(nu))('1/p), 1(LESSTHEQ)p(LESSTHEQ)q<(INFIN). The characterization for the second inequality, even with (omega)=(nu)=(lamda)=Lebesgue measure, leads to a new result in that it characterizes the domains (OMEGA) (L-HOOK EQ) (//R)('n) which support such an inequality. This has been done by C. Amick for the case 1 |