Editing Final Report 2011 (section)

===Results of Tests===
'''Hypothesis 1: The code is an initialism of a poem.'''
Statistics were gathered on the number of words in each line (first, second, third, fourth) of each poem. The statistics gathered include the mean number of words in each line, the standard deviation, the maximum number of words in a line and the minimum.  The results categorized by line number in a Rubaiyat poem are shown in the table below, followed by the statistics from the Somerton Man’s code.
<center>'''Table 1: Letters per Line in Rubaiyat Poems'''</center>
{|border="1" cellspacing="0" style="text-align:center; margin: 1em auto 1em auto"
|- style="color:white; background:MidnightBlue; font-weight:bold"
| width="60" | Line || width="60" | Mean || width="60" | Std Dev|| width="60" | Max || width="60" | Min
|-
| style="color:white; font-weight:bold; background:#4682B4" | First || styl" |8.00 || 1.15 || 10 || 5
|-style="background:#DCDCDC"
| style="color:white; font-weight:bold; background:#4682B4" | Second || 7.69 || 1.20 || 10 || 5
|-
| style="color:white; font-weight:bold; background:#4682B4" | Third|| 7.88 || 1.06 || 10 || 5
|-style="background:#DCDCDC"
| style="color:white; font-weight:bold; background:#4682B4" | Fourth|| 7.87 || 1.31 || 10 || 5
|-
|}

<center>'''Table 2: Letters per Line in Code'''</center>
{|border="1" cellspacing="0" style="text-align:center; margin: 1em auto 1em auto"
|- style="color:white; background:MidnightBlue; font-weight:bold"
| width="100" | Line || width="150" | Number of Letters
|-
| style="color:white; font-weight:bold; background:#4682B4" | First || 9
|-style="background:#DCDCDC"
| style="color:white; font-weight:bold; background:#4682B4" | Second || 11
|-
| style="color:white; font-weight:bold; background:#4682B4" | Third|| 11
|-style="background:#DCDCDC"
| style="color:white; font-weight:bold; background:#4682B4" | Fourth|| 13
|-
|}

The important result is the maximum number of words in the poem lines. Each line category has a maximum number of words of 10 across all of the 75 poems contained in the Rubaiyat. However, the code has 11, 11 and 13 letters in its second, third and fourth lines respectively, each over the maximum. These results allow Hypothesis 1 to be ruled out, giving the conclusion that the code is not an initialism of a Rubaiyat poem.
	
'''Hypothesis 2: The code is related to the initial letters of each word, line or poem.'''
Letter frequency data was gathered on the first letter of each poem, of each line and of each word. This data is plotted against average English initial frequencies and the code letter distribution.

[[Image:CombindeInitialPlots.png|650px|center|All, Line and Poem Initials]]
<center>'''Figure 5 - Letter frequency of initial letters in the Rubaiyat of Omar Khayyam'''</center>

A link between poem initials or line initials and the code can be trivially ruled out. There is a ‘G’ in the code but no line or poem starts with a ‘G’ in the entire Rubaiyat.
A link between all initial letters in the Rubaiyat and the code is more difficult to rule out. There is a generally good correlation between English initials and initials in the Rubaiyat (graphed in light blue) as might be expected, but there are significant discrepancies when compared to the code, such as the code clearly having a greater proportion of A’s, B’s and M’s. While a link cannot be ruled out due to the small sample size of the code (44 letters), for the purposes of this project a link has been ruled unlikely.

'''Hypothesis 3: The code is generally related to the text in the Rubaiyat.'''
This hypothesis was tested by adapting the Java text parser code to generate letter frequency plots for the all letters in the Rubaiyat poems. The results are displayed in the graph below.

[[Image:FullLetterFreqPlot.png|650px|center|All initials]]
<center>'''Figure 6 - Letter frequency of all letters in the Rubaiyat of Omar Khayyam'''</center>

While there is very good correlation between the Rubaiyat poems and English text, the letter frequency of the code is substantially different, with significantly larger proportions of M’s, A’s and B’s. Again the sample size of 44 letters for the code restricts our ability to make a conclusion, but for our purposes there is enough evidence to discount a link.