Experimental conditions of the optical shadow methods of caustics for measurement of the stress intensity factor and the J-integral in various materials(polycarbonate, PMMA, Al 5586D) are investigated. The necessary experimental requirements toe determine accurate values of the stress intensity factors and the J-integrals are described. The ratio of $r_o$ (radius of initial curve) to $r_p$ (plastic zone size) is selected as a parameter to verify the experimental limitation of the method of caustics in determination of fracture parameters. In this study, transmission caustics method was used for compact tension specimens made of polycarbonate and PMMA. while reflection caustics method was applied to c-shaped tension specimen made of Al 5586D. The appropriate ranges of $r_o/r_p$ tp determine accurate values of stress intensity factors were found to be 1.5~1.8. Existing experimental results have been obtained mainly by changing $r_p$ with different loads in $r_o/r_p$. However, in this study we could obtain varying $K_{caus}/K_{th}$ over the wide range of $r_o/r_p$ at fixed load conditions with newly designed optical arrangement. Thus, we could find the range in which theoretical and experimental results agree well each other by changing $r_o$ values only. In Al 5586D specimen, experimental caustics were located inside of the plastic zone, and $K_{caus}/K_{th}$ were found to be not unity in this range. It is found that $J_{caus}/J_{th}=1{;}with{;}r_o/t{geq}0.8$ and the experimental plastic zone includes the contours of caustics.