the 4-digit level of the Standard Industrial Classification (SIC) code." VARIABLE SPECIFICATION Water demand by commercial business establishments is af- fected by a number of different variables, including size of store, hours open per week, nature of water-using appliances, and type of air conditioning system. In addition, as is shown in this study, the price of the product or service sold and the price of water will affect the quantity of water purchased.7 Stated somewhat differently, the quantity of water purchased is "dependent" upon the level of various "independent" variables. Each of the vari- ables studied are defined and discussed in the following. Water Quantity, The Dependent Variable The dependent variable in all models was quantity of water (W) purchased (in thousands of gallons) per month over the period of January 1975 to April 1976. An average figure was necessary because of variations due to broken meters, estimated readings, and illegible or missing observations in meter books or records. Billing periods also varied within and between water companies. This variance was reduced by taking the average over a long period of time. The Price Variable, An Independent Variable The price variable (r) used in all models was the combined sewage and water cost." The price of purchasing water is the relevant price, whether it is "used" for sewage disposal or for other uses. Sewage costs were direct functions of water use in many of the water companies surveyed. The range in price across water companies was $0.30 in Florida City to $3.00 per thou- sand gallons in the Florida Keys. Other Independent Variables Other variables included fixed inputs and prices of outputs. Since these variables differ among business types, a separate dis- "See Appendix A for a further discussion of this matter. ;The conceptual basis for the choice of particular independent variables is discussed in Appendix A. The reader not familiar with economic, derived demand models can review Appendix A to better understand why certain variables were chosen. "Most firms had declining block rate pricing schedules. Price in this study is the marginal price; i.e., the price paid by the firm for the last 1,000 gallons purchased during each month.