440
NATIONAL BUREAU OF STANDARDS NOTES
[J. F. I.
port area to 2} times the throat area. The orifice should also be of sufficient size to give the m a x i m u m gas rate when fully opened and the needle v a l v e used only to reduce the flow of gas. Only after t h e s e adjustments had been made did the burners reach their m a x i m u m h e a t s of 10,000 Btu. per hour for M e k e r t y p e s and 5,000 Btu. for straight-tube types. As a result of this research, several manufacturers are redesigning their burners and the improved appliances will be available in the near future. STABILITY
T E S T FOR ADDITIVE-MOTOR OILS
A new evaluation test for lubricating oils with additives, which simulates the characteristic deposits found in engine tests of m o t o r oils, has been developed in the National Bureau of Standards' lubrication laboratory. The method has several advantages over the present testing procedure, which requires disassembly of the test engine and an examination of its p a r t s . It employs simple laboratory apparatus, is time saving, is less expensive, and eliminates the fuel variable that results from contamination of the oil by the engine fuel. Most important, it gives a quantitative as well as visual rating of an oil. The new test method consists essentially of circulating the oil sample over a heated steel s t r i p in a thin film q u i t e similar to the oil film on an engine cylinder wall. The difference in weight of the s t r i p before and a f t e r the test provides a numerical indication of the a m o u n t of deposit formed. In general, the oils used in the Bureau's investigation are in commercial production and have a considerable background of service performance. They include oils with no additives, premium-type additive oils containing inhibitors and/or detergents, and full heavy-duty oils containing additives. Data obtained with the new test, therefore, give a sensitive indication of the effectiveness of inhibitors, detergents, and their combinations. S A M P L I N G PLAN R E D U C E S INSPECTION C O S T S
Whenever industrial and agricultural products are purchased in q u a n t i t y , the b u y e r must employ some sort of inspection procedure to determine whether or not the material meets the specified standard of quality. The e x t e n t of this inspection will depend on the nature of the material and may r a n g e all the way from a r o u g h appraisal based on an examination of a few i t e m s to the complete inspection of each item. The decision to accept or reject a given lot of material is often determ i n e d by the average quality of a random sample t a k e n from the lot. Whenever such a routine acceptance test is conducted, any way of reducing the a m o u n t of inspection, and therefore the cost, is w o r t h exploring.
Nov., ~949.]
NATIONAL BUREAU OF STANDARDS NOTES
441
The Statistical Engineering laboratory of the National Bureau of Standards has developed a modified double sampling plan that makes possible a saving in the average amount of inspection from 20-40 per cent. In this new plan, the regular testing procedure is interrupted after a predetermined n u m b e r of units have been examined. The material is accepted if the test results at the time of interruption fulfill the criteria of a simple rule that requires neither computation nor the use of tables. If the material does not pass this test, the inspection is completed in the usual manner. In either case, the inspection procedure itself remains unchanged. The modified double sampling plan developed at the National Bureau of Standards is a substitute for the usual single sampling plan that does not a l t e r the sampling itself or require additional computation. In this new plan the same number of units is d r a w n as in the single sample case, but the regular testing procedure is interrupted after a fixed number of items have been tested. If a predetermined number of these test results individually exceed the acceptance value of the original plan, the material is immediately accepted and testing stopped. If less than the required number meet the acceptance value, testing is completed in the usual way and the individual test results are averaged. Acceptance or rejection then depends on whether or not this average exceeds the acceptance value. The chance that material of a given quality will be accepted u n d e r the modified double sampling plan is thus slightly larger than on the original slngle sample plan, since there will be some samples which have the required number of items exceeding the acceptance v a l u e at the point where inspection is interrupted, but which, if completely inspected, would have yielded an overall average below the acceptance value. This increase in the chance of acceptance is desirable when the general quality of the material u n d e r inspection is known to be good. It is undesirable when the quality is bad, and this slight increase in the chance of accepting bad material is the price to be paid for reducing the average amount of inspection without disturbing existing routines. No sampling plan can guarantee the rejection of unserviceable material, but the chance of accepting such material can be made as small as desired by properly choosing the acceptance value and the number of units to be tested.