[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: orion Dating of scrolls (long)

According to Greg Doudna:
> Sigrid Peterson writes:

[Here I'm deleting Greg's quotation of my suggestion to tabulate the
paleogrpahical datings in the DJD Series.]

> I am sorry--I flat out don't accept palaeographical date estimates 
> published in DJD volumes as data. 

See my post to Fred, for a discussion on the four types of data.
Every aggregate of similar statements, or "measures," falls into one of
the four types of data. The data don't care whether you accept them or
not. The facts are friendly, and you can see them or not as you like.

> Those are guesses and based on argument from authority.  Radiocarbon

No, they are judgments based on external referents, "diagnoses," if you
like. The existence of a single authority provides consistency in the
paleographers' assessments, much as does the Diagnostic and Statistical
Manual used by psychologists and psychiatrists. 

 > measurements are data and these are not the same animals.

Both are measurements constituting ordinal data. While the basis for C-14
appears to be interval, the constant rate of decay of a carbon atom, the
actual judgments rendered function as ordinal data.
Sigrid said, based on Greg's reference to the bell curve:
> > probability falls off symmetrically from the top of the bell curve. 

Greg replied:
> This is simply not correct Sigrid. 

Greg, I responded to what you said. You introduced the term "bell-curve."
Apparently I did not respond to what you *meant.  If what you mean is that
C-14 dating probabilities are distributed as a positively skewed
Chi-square distribution of certain parameters, then SAY so. Using a sigma
range, however, implies symmetrical distance along a baseline. Area under
the curve, together with distribution shape, is less misleading. 
I'm deleting a number of further statements about statistical
distributions from this orion post, and will send them to you privately.

Greg said further, about C-14 dating:
>  These dates must be interpreted in the light of being in an archaeological
> assemblage with each other. It is the old problem of outliers in data--what 
> is an outlier, how does one interpret it, what does one do with it. 

One can only make an a priori decision about what data will be collected
and/or analyzed. One cannot look at the data and *then decide what is good
and what is bad.

This is the nub of the struggle between us. You are saying, it seems to
me, something like "Now that I've looked at the C-14/AMS data for
4QpPs(a), I don't like what I see. There must be something wrong, so I
will throw out the data (perhaps with some discussion of how bad it is, so
I feel justified). Without that pesky dating that's out of line, my theory
really works well."

When you do that, your judgment is a posteriori. It's invalid because
you've looked at the data. The scientific attitude says, instead, "The
facts are friendly." They aren't what I expected when I set up my
parameters, but they tell me something I want to know." You can add sotto
voce, "But I didn't know that I wanted to know something new, not like
this!" And then you figure out what's happening, and how you define your
next investigation in a better way, to take into account what you learned.
You refine your measures--of everything, not just the anomalous stuff.
[Greg presents the problem of castor oil contamination.]

You seem to be caught in a circle, saying, "If a date is out of line, then
the problem is probably due to castor-oil contamination. If
there's castor-oil contamination, the date is more recent than it would be
otherwise. The date of 4QpPs(a)  is more recent (than I would like). 
Therefore, there must have been castor-oil contamination. We have come
full circle, and to escape you say, "If the sample was contaminated, then
the results are faulty and I can throw them out.

Where the purpose was to know something, at this point you don't actually
know anything. Instead you are working from presuppositions about unproven
and unprovable contamination. I'm belaboring a point; I'm sure you've
recognized the problem. 

> No, no, no.  It isn't all or nothing, the same with any archaeology.  
> This is the same argument creationists use to disprove the theory of 
> evolution, i.e. show a scientist made and corrected a mistake at some 
> point in the past.

Greg, I am not saying that. I am saying that scientific method is a Method
because there are specific rules which govern the way an investigator
gathers data and analyzes that data. Scientific method is designed to rule
out what have appeared to Steve Goranson and myself as arbitrary decisions
on your part, decisions from authority, your authority, about what is
"good" data, that you will accept, and what is anomalous data that you
disdain to use. You are communicating to us that you are not using
scientific method, you are making "expert" judgments and arguing more or
less from your own authority.

I think you're right; some problems in humanistic investigation have to be
solved by arbitrary decisions. Ideally these are simple decisions that
circumscribe a topic, akin to the a priori decisions about data
collection that the scientist makes. Sometimes you have to solve more
complicated problems by fiat, and have to learn when your solutions are
justified, based on what you know. Just don't call it SCIENCE.

  Radiocarbon dating is proven in principle, the 
> best of the modern labs know what they are doing, and it is a 
> technical issue to beat the contamination spectre with the scrolls.  

The proving of radiocarbon dating is similar in principle to the proving
of paleographical dating. 

> Greg Doudna
> Copenhagen

Sigrid Peterson  UPenn  petersig@ccat.sas.upenn.edu