the neutral approach. We must enforce the law impartially and not become consumer advocates. In turn, we cannot lobby for industry. Both parties require a “third man” to mediate the buying and selling process. You may decide that new programs in the consumer area should include new car warranties that relate to odometer accuracy, tire warranties, accuracy of household measures in such products as carpeting, draperies, zippers, golf balls, and clothing sizes. Inquiries of the "hot line"variety should be referred to consumer associations, better business bureaus, and similar organizations. Concerning the extensive list of weighing and measuring devices, can you test them all annually? Impossible with limited funds, personnel, and equipment. There are several options: 1. Selection of certain devices for universal testing. 2. Selective testing of all devices. 3. Selective testing of certain devices and package control. 4. Selective testing, package control, and consumer investigations. The last option presents a good balance if the selection process is carefully conceived. The basis for selection should be the economic impact in the market-the value of the product sold and location of the producer. Products of local manufacturers and packers should be tested so that full compliance is achieved before commodities and devices enter interstate commerce. The cost of testing and its benefits are important considerations. The testing of gasoline pumps is one of the most misunderstood activities. What is the cost of testing a pump? Four to five dollars? Who benefits most, the station owner or purchaser? Think about that. To build a successful program, the responsibility for accuracy of devices must rest with the owner-operators, and the weights and measures laws must permit selective testing, and not restrict testing to universal, annual inspections. Scale company and meter company service personnel can maintain this equipment accurately, or should (they are getting paid for it), so that nine times out of ten the scale or gas pump will be found "correct" when tested by an official. Ninety percent compliance level is not an unrealistic goal for any jurisdiction. You have chosen the classes of devices to be included in your enforcement program, based upon the cost of inspections, benefits, and economic impact, and now the problem arises as to what constitutes an adequate, reliable sample. The sample cannot be selected at random and be meaningful. It must be based on a well-conceived program, with controls and built-in bias! EDP, although not essential, assists greatly in this procedure. In gasoline pumps, for example, the selection process must be based upon manufacture of meter, brand of product, total gallons metered (totalizer readings), and geographic location. Device history will be an important factor after the program has been in existence for a year, two, or three. With this information, you can select the itinerary, determine the duration between tests, and still maintain 90 percent compliance level (if, in fact, that is a realistic target). Resource allocation is now possible in our reorganized program. We have the information at hand to make sound management decisions: How much time, money, personnel, and equipment should be vested in the various selected areas of activity, using the cost and benefit of inspections, and the expected level of compliance as the criteria. Without adequate, reliable information, an essential tool of management, weights and measures managers are faced with the impossible task of providing business, industry, and the general public an equitable climate where buyer and seller are on even ground. The employment of modern management concepts and techniques will enable the administrator to meet the requirements of today's technology, expanding economy, and population. Weights and measures can respond to the need for effective, economical, and flexible programs within government. A MANAGER'S VIEW OF ELECTRONIC DATA by G. E. MATTIMOE, Deputy Director, Division of Weights and Measures, Hawaii Department of Agriculture The Spanish-American philosopher, George Santayana, once said, "Those who do not remember the past are condemned to relive it again." It would be nearly impossible for me not to remember our past, particularly before we acquired a computer to assist in managing our program. I should like to preface my comments by identifying the proper agency and personnel who, at great expense and much personal sacrifice, were the true prime movers behind whatever achievements may have been accomplished in weights and measures in Hawaii. My warmest thanks go to the National Bureau of Standards in general, and specifically to Dave Edgerly of the Office of Weights and Measures and Walter Urban of the Technical Analysis Division, who made our program an operating reality. I say this without detracting one iota from the monumental efforts of our chief metrologist, George Yokotake, and his staff. In 1966 the Hawaii State Legislature transferred the then existing multi-county weights and measures operations to the state. Included was a hopeless polyglot of records, equipment, and different base systems-if such a term may be so loosely used. We inherited five manual record-keeping systems, which were replaced as rapidly as our resources would permit. Constraints incident to being a new state, the lack of trained personnel, and our peculiar position of being situated in the Department of Agriculture for administrative purposes only added to our early frustrations. We soon recognized that the county records were irreconciliable one to another, so we put them in storage and set about taking a new inventory of the state population of commercially used measuring equipment. With these data as a base, we developed what we thought must be the greatest computer program in the world. In reality, all that we had done was to re-invent the wheel or, more factually, the addressograph. We soon modified this into what must have been the most expensive inventory list in the nation. Not quite convinced that we were not the greatest do-it-yourselfers in the business, we continued to make about every mistake possible, including the proposed use of high-speed optical scanning equipment which had not yet been developed. We published our efforts in a document at which, when we presented it to our statewide information service (our computer capability), they took one look and said, "Forget it!" Wounded, we regrouped and discussed our dilemma. We had developed and designed this obviously great computer system, which we had to operate by hand. Certain that the fates were against us, we made our first sound management decision and sought outside help. Even here we did the wrong thing. We enrolled in computer classes and ingested huge amounts of general information. Then back to the printing press, and out came HI-MARK-S, a copyrighted document on the weights and measures program in Hawaii. When our SWIS people rejected it the second time as being too grandiose, we thought about restoring the monarchy. We were saved from banishment into oblivion by enactment of our state odometer statute, which gives us the responsibility for checking 120,000 odometers per year. This law called for an odometer correctness analyzer (which is a sneaky name for a desk-top computer) and provided funds for purchasing. Through this unit, we were able to batch-process enough data to prove the validity of our program. Only then did SWIS come to our rescue. But we had learned a valuable lesson about attempting to circumvent or encroach upon the sacred domain of the computer people. With 27,000 metrological devices in commercial use in Hawaii, our data base represented equipment the manufacture of some of which had long since been discontinued. Generally speaking, if somebody made one, we got it. Such a divergent population caused us to look hard at our desired output. What did we want or need to develop an effective inspectional system that was subjectively unbiased? We decided upon consideration of the following factors: 1. Inspectional requirements. 2. Optimizing internally scheduled inspections. 3. A management information system. 4. Measures of effectiveness. 5. Cost/benefit ratios. To implement this overall plan, we took the following action: We established the inspection requirements of NBS Handbook 44 through adoption of its examination procedure outlines (EPO's). To optimize inspectional scheduling, we analyzed our data and plotted them in the form of histograms. By overlaying each succeeding histogram on the previous one, we were able to observe any general change in our population, or at least to observe a trend if one started. We noticed that monthly, semiannual, and annual inspections did not necessarily reflect any change in the condition of accuracy of the units tested. (These observations were within the additive limits of uncertainty involved.) We felt that we had a clue, and that usage, not time, seemed to be the more important factor of the two; so we amended our statute to eliminate time-frame inspections in favor of inspecting as often as deemed necessary. We were not quite certain of the definition of "as often as deemed necessary," but we had hopes that we could develop one. I say this regarding usage data: It is not always easy to come by, and it sometimes has to be generated in devious ways! Copies of weight certificates is one approach. These might be forwarded to the Director within 72 hours of issue, and keypunching this information to update unit histories is no great problem. (We do this on a limited scale in Hawaii.) Another might be fitting electronic equipment with totalizers which help establish accuracy decay. Gasoline dispensers and meters in general are "naturals" for this usage approach to reinspections. Their built-in totalizers have been staring at us for years just begging to be read. If your jurisdiction was much like ours, you probably just stared back or ignored them entirely. Under our new administrative law, gasoline dispensers must be tested for correctness compliance on a monthly basis by the user. These test results must be submitted to the Director in a monthly compliance report, which also includes the totalizer readings for each pump. In addition, the dealer must include the total amount of gasoline delivered during the same period by individual drop. This information is entered into the individual gasoline dispenser history and the station history, along with the computer-generated X, R. and σ. The computer analyzes the total gallons received versus the total gallons sold, and prints out these data, citing any gain or loss. The computer identifies the dispensers that are within tolerance, that are losing money, or that have progressed far enough along the frequency of use baseline that their probability of falling in the grey area of uncertainty is imminent and should be serviced/inspected. As each new monthly compliance report is received, it is additive to the existing pump(s) file, so the dealer has a running record to date and a yearly record for reconciliation of profit and loss, most of which is actually a serendipity to our record keeping. We continue our practice of random-sampling the total population, and this sampling program has a definite effect upon the validity and quality of the individual operator's test data. Factually, we conduct a perpetual school in proving gasoline pumps through proper use of a prover, and this has probably been our most important single educational program. There is no numbers game necessary for annual reports to the legislature. We have the facts and the consumer gain or loss, either actual if there is one, or prevented if there is not. These concepts are now being expanded to other devices throughout our operation. The statistical approach to optimum inspection would not be reasonably possible without the aid of the computer. After sufficient data is compiled, the "beyond tolerance" or accuracy decay point can be projected with amazing results. Computer-suggested inspectional programs should be tempered by management experience and knowledge, but the use of data processing to establish the parameters for such suggested inspection cannot be overemphasized. Graphically, the statistical basis for our optimum inspection scheduling looks something like our "accuracy decay curve." Generally speaking, each type of metrological device will show a higher failure rate during the early portion of its service life. This is represented as "new equipment failure" on the chart and is rightfully a part of management's area of concern, particularly should we go into selfcertification of device by company representatives. As the metrological device enters the useful accuracy life span, failures that occur during this period are usually not predictable; they randomly happen to happen. The "used equipment failure" portion represents a sharp increase in inaccuracy to the point of illegality and beyond, to total failure. Such a curve may be generated for |