[CH] Re: need some analogies for scoville units

George Nelson (70431.3065@compuserve.com)
Mon, 2 Mar 1998 08:14:22 -0500 

Michael Bailes wrote:

"[The Scoville]  test was a dilution type test which used a panel of five
taste testers to arrive at a rating, which was measured in increments of
100 Scoville units.  Controlled amounts of Chilli were given to tasters. 
Tasters then recorded pungency in multiples of one hundred.   A majority of
three tasters had to agree before a value was assigned to the food."

A nitpick, but an important one.  The language quoted implied the tasters
would themselves assign a heat value based on their individual assessment. 


My impression is that this is not correct.  The method of the Scoville test
is a threshold one.  The procedure was to dilute the chile powder or
preparation, or whatever (remember, Wilber S. was testing content for
medicinal, not culinary, application) until it was just detectable. 
"Detectable" meant that 3 of the 5 panelists said they detected pepper heat
in the preparation at a threshold level. The number of Scoville Units
assigned was the inverse of the dilution to the nearest 100 units.

For instance, suppose  you extracted 1 gram of powder and  made a "stock"
by diluting it to 100 ml.  If this was detected by the panel, you might
dilute the "stock" preparation 1 in 2, 1 in 4 and 1 in 8.  Supposing the
scoring went like so:

        Preparation     Panelists Detecting heat

        1 in 2                  all
        1 in 4                  3 of 5
        1 in 8                  1 of 5

The dilution required for threshold was (1/100) x (1/2) or 1/200.  The
original powder would then be assigned 200 Scoville units.  Sharp eyes
might detect the uncertainty introduced by not having tried 1 in 5, 1 in 6
or 1 in 7.  A good application of the test would include these, and
illustrates how applying the procedure might become tedious if heat levels
were not known fairly well beforehand.

However, there are a number of important implications of this methodology
as opposed to the one implied in the quote.  First and foremost, it simply
uses the sense of taste as a detector and works at the threshold.  The
question asked about each sample is "Can you detect anything?" rather than
"How hot do you think this is?"  It is therefore "digital" (yes or no)
rather than "analog" (around xx## units).

The trick is to make the right dilutions to present to the panel.

Panelists can be minimally trained, or receive no training at all. 
Finally, they do not extinguish as rapidly because their exposure is always
to very low levels of capsaicinoids.  It is not a "torture test" for the
panelists (not very much fun for a Chile-Head).  It is time-consuming and
tricky since panelists should "rest" after two or at most three successful
detections.  To be efficient,  you have to get the dilutions nearly correct
each time.

ASTM 1083 is a "shortcut" that does ask the panelists to rate the heat by
placing a mark on a graduated scale after doing a taste comparison with a
known dilution of an N-vanillyl-n-nonamide standard.  Since exposure levels
are higher, no more than two determinations are to be made in one day for
this one.  ASTM claims good correlation with HPLC (r = 0.94).  Again,
Chile-Heads need not apply as tasters.

George Nelson
70431.3065@compuserve.com