\section{1973}
\subsection{Lawson}
"Numerical Correlation and Evaluation in the Comparison of Evidentiary Materials"

Lawson mentions the use of emission spectroscopy to identify glass.

This paper provides a sample of ten automobile head lamp glass fragments in Table 6.




2)sets of known distributions for those attributes (weak)

page 5 talks about ten samples of auto head lamp glass (table 6)

3)forensic decision procedures based on glass fragments (strong)

A bunch of statistical equations were thrown together to develop a technique for:
   1. Sorting the results or getting the smallest numerical index so that the closest match of

the parameters involved can be easily evaluated.

   2. Use of the technique to establish a threshold or level of reproducibility of the par-

ticular technique and material involved.


Note: This seems like it is probably useful

@ARTICLE{Seheult,
   author = "Allan Seheult",

   title = "On a Problem in Forensic Science",

   date = 19781200,

   }


1)lists of attributes known for glass (medium)



2)sets of known distributions for those attributes (weak)



3)forensic decision procedures based on glass fragments (strong)


Note: This article summarized : “A Neyman-Pearson test of identification in forensic science is shown to reflect all the properties of a Bayes factor approach presented by D.V. Lindley in a previous volume of this journal.”
 Basically, this article just shows the application of an algorithm for statistical analysis of RIs ( and that the analysis should take into account the distribution of RIs of window glass.


@ARTICLE{Grove,
   author = "D.M. Grove",

   title = "Interpretation of Forensic Evidence Using a Likelihood Ratio",

   date = 19800400,

   }

1)lists of attributes known for glass (medium)



2)sets of known distributions for those attributes (weak)



3)forensic decision procedures based on glass fragments (strong)


Note: This is another paper about algorithms, specifically for determining guilty/not guilty. It's far too deep to for me to look over in a few minutes and understand, but it'll obviously be a very important article for further examination.

@ARTICLE{Evett2,
   author = "I.W. Evett",
   title = "A Quantitative Theory for Interpreting Transfer Evidence in Criminal Cases",
   date = 19840000,
   }

1)lists of attributes known for glass (medium)



2)sets of known distributions for those attributes (weak)



3)forensic decision procedures based on glass fragments (strong)
Somewhat complicated algorithm to calculate the likelihood ratio that contact between two bodies has been made (eg window-suspect, victim-suspect)
@ARTICLE{Evett,
   author = "IW Evett",
   title = "A Bayesian Approach to the Problem of Interpreting Glass Evidence in Forensic Science Casework",
   date = 19840927,
   }

1)lists of attributes known for glass (medium)



2)sets of known distributions for those attributes (weak)



3)forensic decision procedures based on glass fragments (strong)
One again Evett’s article contains another lengthy algorithm and the discussion of it, which again is probably too deep to comprehend in a short time.


@ARTICLE{Evett,
   author = "Ian W Evett and John Buckleton",

   title = "The Interpretation of Glass Evidence. A Practical Approach",

   date = 19900618,

   }


1)lists of attributes known for glass (medium)



2)sets of known distributions for those attributes (weak)
A couple graphs on the distribution of glass on random individuals not known to be connected to a crime, which provides a (perhaps) useful amount of distribution data.


3)forensic decision procedures based on glass fragments (strong)
Evett's at it again, though this time he presents a very simple method for determining the likelihood ratio (eg of the suspect being the one who broke the window) 

4) Conclusions

(note to myself: I've already copied the useful pictures and such to be thrown into the “glass knowledge” paper)


@ARTICLE{Evett,
   author = "IW Evett, JA Lambert, JS Buckleton",

   title = "Further observations on glass evidence interpretation",

   date = 19940708,

   }



1)lists of attributes known for glass (medium)



2)sets of known distributions for those attributes (weak)



3)forensic decision procedures based on glass fragments (strong)
another Evett paper, another algorithm. This one expands slightly on previous work, and introduces one new element: the suspect saying/not saying he was near another glass object after analysis has already been done.


4) Conclusions

@ARTICLE{Walsh,
   author = "KAJ Walsh, JS Buckleton, CM Triggs",

   title = "A Practical example of the interpretation of glass evidence",

   date = 19940801,

   }


1)lists of attributes known for glass (medium)



2)sets of known distributions for those attributes (weak)



3)forensic decision procedures based on glass fragments (strong)
The equation presented in this article is similar to one of Evett's. The difference is that Walsh argues that instead of incorporating grouping and matching, only grouping should be included. Walsh says this is because match/non-match is really just an arbitrary line.

4) Conclusions


@ARTICLE{Koons,
   author = "Robert D. Koons, JoAnn Buscaglia",

   title = "The Forensic Significance of Glass Composition and Refractive Index Measurements",

   date = 19990000,

   } 


1)lists of attributes known for glass (medium)
RI (at D line), Ca, Fe, Al, Mn, Sr, Mg, Ba, Ti, Zr, Na


2)sets of known distributions for those attributes (weak)
graphs of elemental composition, but no tables


3)forensic decision procedures based on glass fragments (strong)
By comparing all of the above attributes, it is extraordinarily unlikely to find two unrelated pieces of glass that match – between 10^-5 and 10^-13 (this would be the denominator in the likelihood ratio

4) Conclusions

@ARTICLE{Koons,
   author = "Robert D. Koons, JoAnn Buscaglia",

   title = "Interpretation of Glass Composition Measurements: The Effects of Match Criteria on Discrimination Capability",

   date = 20020000,

   }

--matt