74 SOCIAL SCIENTIST same, some of these limitations are being overcome by incorporating substitution effect in our studies which are now in progress. The NSS reports also give the distribution of persons by thirteen per capita expenditure classes separately for rural and urban areas. We have estimated from them the parameters of the log-normal distribution by using quantile method. The estimates are presnted in table A-3. TABLE A-3 ESTIMATES OF (P-, 7) Parameter Estimate Rural Urban P. 2.8699 3.1957 \ 0.6465 0.6885 By using parameter estimates given in table A-3, the average xpenditure for each decile class has been computed by using (7) and (8) and the estimates are presented in table A-4. ^ FABLE A-4 ESTIMATED AVE: RAGE TOTAL EXPENDITURE Decile Class I Estimated Average Rural (Rs) 5.85 Total Expenditure Urban (Rs) 7.60 II III 9.02 11.39 11.9815.38 IV 13.75 18.75 V 16.32 22.49 VI 19.12 26.63 VII 22.64 31.88 VIII 27.36 38.89 IX 34.76 50.16 x 57.08 85.13 Using the estimated expenditures, the cost of living indices for each decile class have been arrived at (See tables IV & V.) We have already referred to some limitations of the data. In addition the study is restricted by the presence of a non-monetized economy in the rural areas. These limitations influence the strict accuracy of the estimates. But their broad directions remain unaffected. (The authors acknowledge the help extended by G V S N Murthy, M L Jhala and and L Tombi Singh) 1 D R Gadgil, ^Planning without a Policy Frame", Economic and Political Weekly, Annual Number 1967. 2 The log-normality assumption is fairly well established for the consumer expenditure data. See N S lyengar, ^Analysis of Consumer Expenditure and Data", in A K Das» gupta (Ed) Methodology of Economic Research, Asia Publishing House 1968.