Food Protection Trends - July/August 2017 - 281

Figure 8. Temperature range chart - 3 sigma lower limit - using d2 calculations

core temperature data can be graphed by floor personnel by
calculating the adjusted cumulative frequency and plotting
it against the individual core temperatures on normal
probability paper (2, 7). For an example of a variable tally
sheet that displays a histogram of sample data, see Appendix
C-1. This tally sheet also includes two sets of statistics: a)
the average plus the standard deviation, and b) the median
and the range/d2. The normal probability plot of cumulative
data from Appendix C-1 is plotted in Appendix C-2; since the
data are nearly linear, the distribution can be assumed to
be normal or near-normal. Several authors use the term "fat
pencil test" to test for normality or near normality. If a fat
pencil covers the spots on a normal probability plot, the data
can be assumed to be normal/near-normal (4).
If there are questions on normality, plant personnel can
plot the cumulative percentages on normal probability
plot paper. Instructions on how to prepare a blank normal
probability chart in an Excel spreadsheet can be found in
Appendix C-3. The calculation and plotting can be done
quickly by hand or with calculators in the precooking
office. These normality plots with the precooking core
temperature records are to be kept for HACCP verification
during record review.
Variable sampling plans
Either variable sampling plan is a quick test if there
are enough thermometers to use during the monitoring
procedure. The tally sheet collection system is a simple
yet very powerful technique. A tally sheet printed on
waterproof paper (37) with 1-degree frequency bins will
work very well. Waterproof paper is suggested because of
the steamy environment around the precookers. The data

can be quickly gathered, tallied, and analyzed with multiple
fast-acting thermometers and a well-designed tally sheet.
The average and standard deviation method (i.e.,
the computer method) is preferred for sample size
considerations. The calculations can be made with the
grouped data tally sheet method and a spreadsheet, as is
shown in Appendix C-1.
The median
and range/d2 test for the minimum core
temperature can be quickly and easily calculated from the
collected and grouped data. The statistics can be calculated
on the spot with simple counting techniques and a bit of
division (Appendix C-1). The [Median - (3 Range/d2)]
will indicate the lower limit of the distribution of core
temperatures, and a decision can be made whether to:
(a) release, (b) measure again and release, or (c) rework
with more precooking. With regard to the decision to
wait and measure again, the decision can be based on
whether the lower limit is below the lower specification
(needs rework), between the lower specification and a
predetermined marginal specification (wait) or above the
marginal specification (release). For an example, see Appendix
C-1. One can be X% confident that at least Y% of the core
temperatures are above this lower limit. X and Y are indicated
in Table 8 based on the sample size. A visual inspection of the
tally sheet can confirm a normal/near-normal distribution. If
further verification of normality/near-normality is needed,
the precooker operator can plot the cumulative relative
frequency on normal probability plot paper (Appendix C-3).
Attribute sampling plans
From an operational point of view, the use of attribute
sampling is unlikely to be chosen because of the impact on

July/August Food Protection Trends

281
Table of Contents for the Digital Edition of Food Protection Trends - July/August 2017
