Having defined the index profiles for the Chiripa phases, it remained
only to produce an algorithm to find the closest match to specific
assemblage profiles from among the 5151 possible combinations of the
three index profiles. To this end I wrote a simple program in Perl,
reproduced as Appendix C. This
program presents a simple interface in which the user enters the details
of an assemblage composition. The program then constructs an appropriate
assemblage profile and compares this to the 5151 possible combinations
of the three index profiles. It reports all matches deviating from
the assemblage profile by less than 3 percentage points. Finally,
it reports the closest possible combination, as a percentage of the
three Chiripa phases and the number of sherds from the assemblage
which would pertain to each phase. The output from a sample run of
the program is given below. The example given is from surface collection
locus 4339, a 50 m
collection unit at the site of T-394 (Janko
Kala) in the community of Zapana.
[/home/inti]> perl frequency_profile.pl
Enter Paste Group counts in the assemblage to be analyzed:
Paste Group 1: 61
Paste Group 2: 37
Paste Group 3: 21
All others: 31
Summary of data:
Group 1 Group 2 Group 3 Other
Count 61 37 21 31
% of Total 0.41 0.25 0.14 0.21
% of three 0.51 0.31 0.18
Press ENTER to continue
Matches with error less than or equal to 1:
LC% MC% EC% error LC MC EC
71 21 8 0.9 107 32 12
71 22 7 0.7 107 33 11
71 23 6 0.7 107 35 9
71 24 5 0.8 107 36 8
72 20 8 0.9 108 30 12
72 21 7 0.6 108 32 11
72 22 6 0.3 108 33 9
72 23 5 0.4 108 35 8
72 24 4 0.7 108 36 6
73 19 8 1 110 29 12
73 20 7 0.6 110 30 11
73 21 6 0.3 110 32 9
73 22 5 0.2 110 33 8
73 23 4 0.5 110 35 6
73 24 3 0.9 110 36 5
74 20 6 0.7 111 30 9
74 21 5 0.6 111 32 8
74 22 4 0.6 111 33 6
74 23 3 0.9 111 35 5
75 20 5 1 113 30 8
Best match:
LC% MC% EC% error LC MC EC
73 22 5 0.2 110 33 8
Would you like to process another surface collection unit? (y/n)no
Done.
Press Enter to Exit
[/home/inti]>
In this case the best match was a mixture of 73% Late Chiripa, 22%
Middle Chiripa and 5% early Chiripa. The numbers on the right are
the number of actual sherds in the unit from each phase, assuming
that the best match actually represents the given mix of phases. Given
my earlier definition of a site (0.1 sherds/m) and the dimensions
of the circle (50m), I conclude that this particular unit lies
within the boundaries of Early, Middle and Late Chiripa occupations
(sectors) of the site - though the Early Chiripa occupation in this
unit is tenuous. In fact, this unit is from near the edge of the Early
Chiripa sector. This is the method by which I defined occupations
and sectors of the Early and Middle Formative Periods - the Early,
Middle and Late Chiripa phases.
The method outlined in the foregoing is rather labor-intensive and not as precise as I would like. However, at present it is the only method ever proposed for distinguishing occupations of the various Chiripa phases using mixed surface assemblages. I have described it in such detail in the hopes that others may build on this work and hopefully supersede it.
As a final note, I should add that it is very difficult to learn to distinguish the various paste groups as I have described them here. They really cannot be conveyed satisfactorily in writing. I strongly urge anyone planning to employ this method to obtain a reference collection of unmixed Chiripa phase assemblages and to develop their own index profiles for the phases, as I did. The index profiles I have presented above will not be precisely valid for anyone but myself.