If you search for Genesis 1:1 in long strings created from its characters
offset by digits of 
'PI' + 'e', 
'PI' + 'PHI', and 
'PI' + 'e' + 'PHI' 
you find the words more quickly 
than when doing the same with the words
scattered. 

This was tested against one hundred thousand random strings.

This is totally ignored in our one in google calculation.
'PI'
by itself is random and 
'PI' + random digits 
gives a negative coorelation. 'piephi.txt' also uses a nonsense sentence
to show a normal distribution. Feel free to substitute your own sentence.
To disprove this theory, on top of everything else, the sentence
must give 441 (3 X 7 X 7 X 3) or more out of 510 for 'PI' + 'e', 'PI' + 'PHI',
and 'PI' + 'e' + 'PHI'.
&nbsp &nbsp &nbsp
<a href='piephi.java'>piephi.java</a>
&nbsp &nbsp &nbsp
<a href='runpiephi.java'>runpiephi.java</a>
&nbsp &nbsp &nbsp
<a href='piephi.txt'>piephi.txt</a>
&nbsp &nbsp &nbsp
<a href='myout.txt'>myout.srt</a>