Thanks, John, for correcting the false impression that I gave that the MAV is derived *directly* from the random number tables. As you state, the random numbers are used only to determine which packages are chosen for inspection from the larger total lot of packages available to the inspector.
A limited number of packages may have fill errors below the MAV. Quotation: "Each sampling plan limits the number of negative package errors permitted to be greater than the MAV." which disputes "No individual shortfall..." Gene. ---- Original message ---- >Date: Sun, 27 Mar 2011 13:13:23 -0700 (PDT) >From: "John M. Steele" <[email protected]> >Subject: Re: [USMA:50212] Re: Williams Changes Their Label -- I had an impact >To: "U.S. Metric Association" <[email protected]>, [email protected] > >There are two requirements generally, given in simple form in section 1.2 >(with certain >exceptions): > >*The average of the sample must equal or exceed the declared contents. > >*No individual shortfall (from declared) may exceed MAV. The MAVs come from >lookup tables >based on mass, volume, etc. Frankly, the requirements for uniformity don't >seem very >stringent. > > > >The random number table is used only to "randomly" pick the samples, not for >any >compliance calculation. I assume the test for average is based on the >Student-t test, but >the details aren't given. Unless you really enjoy statistics, the details get >a little >complex. > >--- On Sun, 3/27/11, [email protected] <[email protected]> wrote: > > > > From: [email protected] <[email protected]> > > Subject: [USMA:50212] Re: Williams Changes Their Label -- I had an impact > > To: "U.S. Metric Association" <[email protected]> > > Date: Sunday, March 27, 2011, 3:43 PM > > > > The testing of packages for compliance with legal marketing requirements is > based on the > idea of "Maximum Ållowable Variation (MAV) as follows: > > > > 1. An Inspector selects random samples from a lot (a shipment) of the > product. > > > 2. The amount in each package from the samples is measured and tabulated. > > > > 3. The Inspector determines the MAV from a Random Number Table. > > > > 4. The Inspector evaluates compliance with the MAV requirement. > > > > 5. The Inspector reports deviations from the MAV for enforcement actions. > > > > Documentation is in NIST Handbook 133, Chapter 2. "Basic Test Procedure" > (also a free > download). Numerical examples of Sample Size, Measured Values (some over and > some under > the "declared value"), and MAV are in the Appendix of Handbook 133. Is this > the > explanation you want? EAM > > > > ---- Original message ---- > > >Date: Sun, 27 Mar 2011 14:31:33 -0400 > > >From: "Kilopascal" <[email protected]> > > >Subject: [USMA:50210] Re: Williams Changes Their Label -- I had an impact > > >To: <[email protected]>, "U.S. Metric Association" <[email protected]> > > > > > > Seeing that the document is 298 pages in the PDF > > > format, can you direct me to the page where it > > > states that the contents of a package may be less > > > than what is declared? > > > > > > > http://ts.nist.gov/WeightsAndMeasures/Publications/upload/HB130-2009-PDF-FullDoc.pdf > > > > > > I'm not sure what you refer to as "some" and what > > > their exceptions may be. > > > > >[USMA:50210] Re: Williams Changes Their Label -- I had an > > > impact > > > > > > mechtly > > > Sun, 27 Mar 2011 10:57:47 -0700 > > > > > > "Kilopascal" <[email protected]> claimed: > > > ... > > > > in the US, the > > > > declarations represent minimum quantities. > > > > Everything is going to be a few grams or more over > > > > stated value. > > > > > > Not True! Some packages may be *under* the amount stated on the label. > > > See NIST Handbook 130 (a free download). >
