For each basic model of metal halide lamp ballast selected for testing, a sample of sufficient size, no less than four, shall be selected at random and tested to ensure that:
(a)
Any represented value of estimated energy efficiency calculated as the measured output power to the lamp divided by the measured input power to the ballast (Pout /Pin ), of a basic model is no less than the higher of:
(1)
The mean of the sample, or
(2)
The upper 99-percent confidence limit of the true mean divided by 1.01. (b) Any represented value of the energy efficiency of a basic model is no greater than the lower of:
(1)
The mean of the sample, or
(2)
The lower 99-percent confidence limit of the true mean divided by 0.99.
Code of Federal Regulations
[75 FR 10968, Mar. 9, 2010]