Frank＇s　theory　describes　that　a　screw　dislocation　will　produce　a　pit　on　the　surface,　and　has　been　evidenced　in　many　material　systems　including　GaN.　However,　the　size　of　the　pit　calculated　from　the　theory　deviates　significantly　from　experimental　result.　Through　a　careful　observation　of　the　variations　of　surface　pits　and　local　surface　morphology　with　growing　temperature　and　V/III　ratio　for　c-plane　GaN,　we　believe　that　Frank＇s　model　is　valid　only　in　a　small　local　surface　area　where　thermodynamic　equilibrium　state　can　be　assumed　to　stay　the　same.　If　the　kinetic　process　is　too　vigorous　or　too　slow　to　reach　a　balance,　the　local　equilibrium　range　will　be　too　small　for　the　center　and　edge　of　the　screw　dislocation　spiral　to　be　kept　in　the　same　equilibrium　state.　When　the　curvature　at　the　center　of　the　dislocation　core　reaches　the　critical　value　1/r(0),　at　the　edge　of　the　spiral,　the　accelerating　rate　of　the　curvature　may　not　fall　to　zero,　so　the　pit　cannot　reach　a　stationary　shape　and　will　keep　enlarging　under　the　control　of　minimization　of　surface　energy　to　result　in　a　large-sized　surface　pit.