Info on The Measurements

T/E Ratio - legal ratio 4:1 or less
13C/12C - a normal ratio -21.3 to -24.4/1000 (delta 13C/1000 range)
Typical user ratio -27.43 +/- 0.76/1000 (mean and SD).

Following on from a previous discussion on the accuracy of the testing,
the 13C/12C ratio is very sensitive to error of measured values. This
isn't a case of categorical data (it's either synthetic or it isn't);
the determination is made from what should be very precise quantitive
data.

We know what Floyd's T/E ratio is; I've seen all the reports and it
ranges from 5:1 to 11:1 (LOL). I do hope we get a reliable report on his
delta 13C value.

Phil H

Phil Holman wrote:
T/E Ratio - legal ratio 4:1 or less
13C/12C - a normal ratio -21.3 to -24.4/1000 (delta 13C/1000 range)
Typical user ratio -27.43 +/- 0.76/1000 (mean and SD).

Following on from a previous discussion on the accuracy of the testing,
the 13C/12C ratio is very sensitive to error of measured values. This
isn't a case of categorical data (it's either synthetic or it isn't);
the determination is made from what should be very precise quantitive
data.

We know what Floyd's T/E ratio is; I've seen all the reports and it
ranges from 5:1 to 11:1 (LOL). I do hope we get a reliable report on his
delta 13C value.

I lack the informational base to understand what you are saying above.
A ratio of two positive counts can't be negative, so I'm going to
assume that's the log of the ratio. I'm going to try to put what you
said above in different terms, so you can see where I'm likely lost.

Let's suppose there are 10,000,000,000,000 carbon 12 atoms observed
(1*10^13).

If we observed 5617 carbon 13 atoms, this would be a ratio of 5617/
10,000,000,000,000, which is a tiny number. The natural log of this
ratio would be -21.3.

If we observed 253 carbon 13 atoms, this would be a ratio of 253/
10,000,000,000,000, which is a tiny number. The natural log of this
ratio would be -24.4.

In the typical drug case, there is a ratio of -27.43 which would be 12
carbon13 atoms out of 10,000,000,000,000.

If we go down 3 standard deviations on this ratio (-27.43 +3*.76 =
-25.15), this would be 119 atoms.

Since I didn't make use of the "/1000" above I think I've misunderstood
something. Also, this sounds like a lot of carbon atoms for a urine
sample that's mostly water, but I'm again ignorant.

It's also not clear from your post whether the measurement error is
already factored into the normal range. In other words, is that
normal ratio based on "normal measurements given this amount of urine
tested under these conditions" -- in which case the applicable
measurement error is already factored in -- or is it the normal ratio
based on idealized information, in which case measurement error would
be an additional source of error?

This isn't a criticism; just wondering. This isn't stuff I ever
figured I'd want/need to know about.

"Bob Dole" wrote in message
oups.com...
Phil Holman wrote:
T/E Ratio - legal ratio 4:1 or less
13C/12C - a normal ratio -21.3 to -24.4/1000 (delta 13C/1000 range)
Typical user ratio -27.43 +/- 0.76/1000 (mean and SD).

Following on from a previous discussion on the accuracy of the
testing,
the 13C/12C ratio is very sensitive to error of measured values. This
isn't a case of categorical data (it's either synthetic or it isn't);
the determination is made from what should be very precise quantitive
data.

We know what Floyd's T/E ratio is; I've seen all the reports and it
ranges from 5:1 to 11:1 (LOL). I do hope we get a reliable report on
his
delta 13C value.

I lack the informational base to understand what you are saying above.
A ratio of two positive counts can't be negative, so I'm going to
assume that's the log of the ratio. I'm going to try to put what you
said above in different terms, so you can see where I'm likely lost.

Let's suppose there are 10,000,000,000,000 carbon 12 atoms observed
(1*10^13).

If we observed 5617 carbon 13 atoms, this would be a ratio of 5617/
10,000,000,000,000, which is a tiny number. The natural log of this
ratio would be -21.3.

If we observed 253 carbon 13 atoms, this would be a ratio of 253/
10,000,000,000,000, which is a tiny number. The natural log of this
ratio would be -24.4.

In the typical drug case, there is a ratio of -27.43 which would be 12
carbon13 atoms out of 10,000,000,000,000.

If we go down 3 standard deviations on this ratio (-27.43 +3*.76 =
-25.15), this would be 119 atoms.

Since I didn't make use of the "/1000" above I think I've
misunderstood
something. Also, this sounds like a lot of carbon atoms for a urine
sample that's mostly water, but I'm again ignorant.

A delta 13C/1000 value is used which I take as converting to a range of
975.6/1000 to 978.7/1000 for a normal 13C/12C ratio [-21.3 to -24.4/1000
(delta 13C/1000 range)]
A "user" ratio would be 972.57/1000 +/- 0.76 which means that approx
2/3rds of the subjects tested fell within the range of 971.81/1000 to
973.33/1000.


It's also not clear from your post whether the measurement error is
already factored into the normal range. In other words, is that
normal ratio based on "normal measurements given this amount of urine
tested under these conditions" -- in which case the applicable
measurement error is already factored in -- or is it the normal ratio
based on idealized information, in which case measurement error would
be an additional source of error?

Any kind of distribution will include some error in measurement but
nothing like the 20% error mentioned in a previous post. See the
"Testosterone Testing: More on False Positives" thread (7/29/06). We are
looking at a difference as small as 3/1000 to determine either a
positive or negative result which means it is very sensitive to error.


This isn't a criticism; just wondering. This isn't stuff I ever
figured I'd want/need to know about.


Phil H

"Phil Holman" piholmanc@yourservice wrote in message
...

"Bob Dole" wrote in message
oups.com...
Phil Holman wrote:
T/E Ratio - legal ratio 4:1 or less
13C/12C - a normal ratio -21.3 to -24.4/1000 (delta 13C/1000 range)
Typical user ratio -27.43 +/- 0.76/1000 (mean and SD).

Following on from a previous discussion on the accuracy of the
testing,
the 13C/12C ratio is very sensitive to error of measured values.
This
isn't a case of categorical data (it's either synthetic or it
isn't);
the determination is made from what should be very precise
quantitive
data.

We know what Floyd's T/E ratio is; I've seen all the reports and it
ranges from 5:1 to 11:1 (LOL). I do hope we get a reliable report on
his
delta 13C value.

I lack the informational base to understand what you are saying
above.
A ratio of two positive counts can't be negative, so I'm going to
assume that's the log of the ratio. I'm going to try to put what you
said above in different terms, so you can see where I'm likely lost.

Let's suppose there are 10,000,000,000,000 carbon 12 atoms
observed
(1*10^13).

If we observed 5617 carbon 13 atoms, this would be a ratio of 5617/
10,000,000,000,000, which is a tiny number. The natural log of this
ratio would be -21.3.

If we observed 253 carbon 13 atoms, this would be a ratio of 253/
10,000,000,000,000, which is a tiny number. The natural log of this
ratio would be -24.4.

In the typical drug case, there is a ratio of -27.43 which would be
12
carbon13 atoms out of 10,000,000,000,000.

If we go down 3 standard deviations on this ratio (-27.43 +3*.76 =
-25.15), this would be 119 atoms.

Since I didn't make use of the "/1000" above I think I've
misunderstood
something. Also, this sounds like a lot of carbon atoms for a urine
sample that's mostly water, but I'm again ignorant.

A delta 13C/1000 value is used which I take as converting to a range
of 975.6/1000 to 978.7/1000 for a normal 13C/12C ratio [-21.3
to -24.4/1000 (delta 13C/1000 range)]
A "user" ratio would be 972.57/1000 +/- 0.76 which means that approx
2/3rds of the subjects tested fell within the range of 971.81/1000 to
973.33/1000.

Well, not exactly. The delta value is correctly explained in this
article.

http://www.brookscole.com/chemistry_...s/ch01-01.html

dx sample(per thousand) = [(Rsample - Rstandard) / Rstandard] x 1000


Phil H

"Bob Dole" wrote in message
oups.com...
Phil Holman wrote:

in response to questions I had about how to interpret the numbers he
posted

Well, not exactly. The delta value is correctly explained in this
article.

http://www.brookscole.com/chemistry_...s/ch01-01.html

dx sample(per thousand) = [(Rsample - Rstandard) / Rstandard] x 1000


Thanks. I see where I went off-track now.


Having probably confused the issue I'll summarize:
Given a standard ratio of a 13C:12C isotopic composition, natural
testosterone has approx 23 less parts per thousand of 13C. Synthetically
enhanced testosterone has approx 27.5 less parts per thousand of 13C.
Taking into account normal distributions around these values, a 10%
error in measurement could produce a bogus result.

I'll throw some numbers in to test the outcome.
Say our reference standard has a 13C/12C ratio of 1.11/98.89
and our testosterone sample has a ratio of 1.08/98.92
This will result in a delta 13C/1000 of [(1.08/98.92 - 1.11/98.89) /
1.11/98.89] x 1000
= -27.3

Phil H

Phil Holman wrote:
"Bob Dole" wrote in message
oups.com...
Phil Holman wrote:
T/E Ratio - legal ratio 4:1 or less
13C/12C - a normal ratio -21.3 to -24.4/1000 (delta 13C/1000 range)
Typical user ratio -27.43 +/- 0.76/1000 (mean and SD).

Following on from a previous discussion on the accuracy of the
testing,
the 13C/12C ratio is very sensitive to error of measured values. This
isn't a case of categorical data (it's either synthetic or it isn't);
the determination is made from what should be very precise quantitive
data.

We know what Floyd's T/E ratio is; I've seen all the reports and it
ranges from 5:1 to 11:1 (LOL). I do hope we get a reliable report on
his
delta 13C value.

I lack the informational base to understand what you are saying above.
A ratio of two positive counts can't be negative, so I'm going to
assume that's the log of the ratio. I'm going to try to put what you
said above in different terms, so you can see where I'm likely lost.

Let's suppose there are 10,000,000,000,000 carbon 12 atoms observed
(1*10^13).

If we observed 5617 carbon 13 atoms, this would be a ratio of 5617/
10,000,000,000,000, which is a tiny number. The natural log of this
ratio would be -21.3.

If we observed 253 carbon 13 atoms, this would be a ratio of 253/
10,000,000,000,000, which is a tiny number. The natural log of this
ratio would be -24.4.

In the typical drug case, there is a ratio of -27.43 which would be 12
carbon13 atoms out of 10,000,000,000,000.

If we go down 3 standard deviations on this ratio (-27.43 +3*.76 =
-25.15), this would be 119 atoms.

Since I didn't make use of the "/1000" above I think I've
misunderstood
something. Also, this sounds like a lot of carbon atoms for a urine
sample that's mostly water, but I'm again ignorant.

A delta 13C/1000 value is used which I take as converting to a range of
975.6/1000 to 978.7/1000 for a normal 13C/12C ratio [-21.3 to -24.4/1000
(delta 13C/1000 range)]
A "user" ratio would be 972.57/1000 +/- 0.76 which means that approx
2/3rds of the subjects tested fell within the range of 971.81/1000 to
973.33/1000.


It's also not clear from your post whether the measurement error is
already factored into the normal range. In other words, is that
normal ratio based on "normal measurements given this amount of urine
tested under these conditions" -- in which case the applicable
measurement error is already factored in -- or is it the normal ratio
based on idealized information, in which case measurement error would
be an additional source of error?

Any kind of distribution will include some error in measurement but
nothing like the 20% error mentioned in a previous post. See the
"Testosterone Testing: More on False Positives" thread (7/29/06). We are
looking at a difference as small as 3/1000 to determine either a
positive or negative result which means it is very sensitive to error.


This isn't a criticism; just wondering. This isn't stuff I ever
figured I'd want/need to know about.


Phil H

Does this mean that the representative of mass spectrometer
manufacturer had it right when he pointed out that there are regularly
errors in isotope tests.

"But a more-cautious note was sounded by Dr. Davis, who is now the
technical director for Mass Spec Solutions Ltd., a Wythenshawe, U.K.,
maker of mass-spectrometry devices. "Quite regularly there are errors
in the isotope tests," he said. "It's a very difficult analytical
technique.""

The WSJ article this quote is from concluded:

"The apparent sensitivity of the testosterone test's numbers to alcohol
consumption, and the announcement of partial test results without full
disclosure by the cycling union, has created a milieu for cyclists that
is "almost Kafka-esque," Dr. Davis said. "The phrase often bandied
about is 'chemical McCarthyism'.""

http://online.wsj.com/public/article... main_tff_top

wrote:

Does this mean that the representative of mass spectrometer
manufacturer had it right when he pointed out that there are regularly
errors in isotope tests.

There certainly can be errors, but unless the equipment isn't
functioning properly they are unlikely to arise in the actual
determination of the 13C:12C ratio of the CO2 that is produced and
measured during GC/C/IRMS. A more likely source of error would be in
the purification procedures used to prepare the samples for GC/C/IRMS
analysis (which is why it would be important to have some reference
steroid material with a known 13C:12C to verify any such analyses,
especially for a multi-center "study" such as is the case for drug
testing).

Andy Coggan

Phil Holman wrote:
We know what Floyd's T/E ratio is; I've seen all the reports and it
ranges from 5:1 to 11:1 (LOL). I do hope we get a reliable report on his
delta 13C value.

From a NYTImes article posted to ST... his ratio was 3.99 units higher,
with 3 or greater indicating a positive result.

"The test starts with an isolation of testosterone from the athlete's
urine. Then chemists determine the makeup of the carbon atoms that form
the backbone of testosterone.

Ordinarily, carbon atoms are made up of six protons and six neutrons,
giving them an atomic weight of 12. But occasionally, they have an
extra neutron, giving them an atomic weight of 13.

By chance, soy plants are the source of most pharmaceutical
testosterone. They tend to have slightly less carbon-13 than other
plants that are more abundant in the human diet. Humans make
testosterone from the food they eat, so their testosterone typically
has more carbon-13 than the testosterone that drug companies synthesize
from soy.

But these differences are tiny.

The test determines whether the testosterone in the athlete's urine
has less carbon-13 than another naturally occurring hormone in the
urine, like cholesterol. The test is considered positive when the
carbon isotope ratio - the amount of carbon-13 compared to carbon-12
- is three or more units higher in the athlete's testosterone than
it is in the comparison hormone. It is evidence that the testosterone
in the urine was not made by the athlete's body. Landis's
difference was 3.99, according to his own doctor."

"Ron Ruff" wrote in message
oups.com...

Phil Holman wrote:
We know what Floyd's T/E ratio is; I've seen all the reports and it
ranges from 5:1 to 11:1 (LOL). I do hope we get a reliable report on
his
delta 13C value.

From a NYTImes article posted to ST... his ratio was 3.99 units
higher,
with 3 or greater indicating a positive result.

"The test starts with an isolation of testosterone from the athlete's
urine. Then chemists determine the makeup of the carbon atoms that
form
the backbone of testosterone.

Ordinarily, carbon atoms are made up of six protons and six neutrons,
giving them an atomic weight of 12. But occasionally, they have an
extra neutron, giving them an atomic weight of 13.

By chance, soy plants are the source of most pharmaceutical
testosterone. They tend to have slightly less carbon-13 than other
plants that are more abundant in the human diet. Humans make
testosterone from the food they eat, so their testosterone typically
has more carbon-13 than the testosterone that drug companies
synthesize
from soy.

But these differences are tiny.

The test determines whether the testosterone in the athlete's urine
has less carbon-13 than another naturally occurring hormone in the
urine, like cholesterol. The test is considered positive when the
carbon isotope ratio - the amount of carbon-13 compared to carbon-12
- is three or more units higher in the athlete's testosterone than
it is in the comparison hormone. It is evidence that the testosterone
in the urine was not made by the athlete's body. Landis's
difference was 3.99, according to his own doctor."

Thanks Ron. A negative for the delta 13C value tends to be confusing.
You reported the NYT info and went on to say "the amount of carbon-13
compared to carbon-12
is three or more units higher" Shouldn't this be 3 or more units lower.

Phil H

Phil Holman wrote:
Thanks Ron. A negative for the delta 13C value tends to be confusing.
You reported the NYT info and went on to say "the amount of carbon-13
compared to carbon-12
is three or more units higher" Shouldn't this be 3 or more units lower.

I don't know... I was hope you (or someone else) would know what it
meant.