We investigate the conditions under which the Horvitz-Thompson estimator (HTE) is asymptotically design unbiased (ADU), design consistent (ADC), almost surely (a.s.) convergent and convergent in rth mean under a framework developed by Fuller and Isaki (1981), Isaki and Fuller (1982). We also investigate the rate of con-vergence of HTE in situations where HTE is asymptotically normally distributed. Lastly we prove a central limit theorem for the asymptotic distribution of a - weighted estimator of regression coefficient under simple linear regression model with heteroscedastic error.