Rajendra SinghSuman Kumar2016-12-202016-12-202006http://krishikosh.egranth.ac.in/handle/1/91535Judicious use of fertilizer form one of the most important means of stepping up agricultural production. For many cropping situation, especially in developing countries some models do not fit the responses of certain crops to fertilizer. Performance of quadratic, square root, inverse quadratic and modified quadratic response functions have been examined on major food crops grown in Haryana. Based on R2 criterion, inverse quadratic model gives better fit to Pearlmillet crop, where as quadratic function describes the yield-fertilizer relationship in Rice crop. It has been observed that a modified quadratic function is better suited when data shows initial sharp rise at lower levels of fertilizer application. Fertilizer yield data of wheat is often available of many varieties and/or for a numbers of years. Dummy variables have used for combining data for different varieties and over years and for testing structural stability of quadratic response function for three varieties; HD2329,WH542 and WH533 of wheat crop grown in Haryana. Models for individuals varieties have also been deduced from the pooled regression. A comparative study of quadratic, square root, inverse bilinear, and cobb-douglas response models used for the two nutrients experiment. The canonical form is also performed for quadratic and square root models. The characteristic roots of real symmetric matrix of estimates are measured for the coefficients of quadratic and interaction terms. Based on characteristic roots and R2, square root model give better fit for wheat crop.enQuadratic, Square root, Inverse quadratic, Modified quadratic, Cobb-Douglas, Response surface, Dummy variable, Least square estimate, Economic optimumThesis