A classification for use in statistical testing must “…produce classes that so accurately reflect the nature of the specimens that those classes and criteria that define them can be used to identify additional specimens of the same kind.”
Vitruvius states that: The planning of temples depends upon symmetry…[and] it arises from proportion…[using] a fixed module…by which the method of symmetry is put into practice. For without symmetry and proportion no temple can have a regular plan; that is, it must have an exact proportion worked out after the fashion of the members of a finely-shaped human body.
[Vitruvius Book III,1,1: translation by Morris Hicky Morgan 1914]
Although Vitruvius might state that symmetry and proportion are expressed in the human form, he confuses the Pythagorean concept of harmonics with an arithmetic or geometric unit to be added and subdivided. Thus the “fixed module” about which Vitruvius wrote is only a geometric constant, rather than an ideal number based on Pythagorean harmonic theory. It is these harmonic rules of proportion that eluded Vitruvius, for although he knew of a proportional method for the construction of Doric temples, he was not privy to their specifics.
That has not deterred many who whish to solve his proportional dilemma, the list of which is extensive [Mc Allister 1969; Adams 1973; Winter 1978; Coulton 1975-5]. But those historians who doggedly cling to the Vitruvius model have always come away disappointed. Nevertheless, with every new attempt the desired results have fallen short of producing a substantive proportional theory. One obstacle inhibiting past researchers from discovering any sort of unified set of laws is the limited number of structures available to form their data structure, producing results that work for only a select group of buildings, but fail to embrace all Doric temples. Berve & Gruben , for example, state that the temple of Zeus at Olympia was “…built up from an ‘internal’ basic measurement [of] the interaxial of 16 Doric feet…a proportion from which all other dimensions were derived.” As convincing as this process is for the temple of Zeus, it falls short of defining, or even outlining, a strategy relevant to other Doric temples.
Consequently, modern Doric temple classifications rely on subjective identifications rather than the quantification of specific data. With this defect in Doric Temple classification, a new system is needed. This new system must incorporate a “…significantly recurring clusters of attributes…[that] represent some sort of norm, standard, or mental template in the minds of the…[ancient] craftsman.” [Whallen 1972:16]. I propose these attributes form groups of temples defined and segregated from all other structures by CLASS. This should help determine specific elements of the Doric style, rather than simply listing arbitrary stylistic elements.
So, what criteria should one use to define a Doric temple Class? The most logical method is to group according to a shared set of attributes or common traits. Specimens must be assembled according to shared diagnostic attribute. Certainly religious architecture consisting of colonnaded structures forms the root of such criteria. After that, columns consisting of capitals, smooth architrave or epistyle; frieze and metope, form the remainder of the attributes [Brillian 1972; Robertson 1940; Dinsmoor 1973; Berve & Gruben 1973]. In the case of the Doric Order: that includes temples consisting of columns with a smooth spreading convex echinus separating the column proper from the architrave above it; columns may or may not have arris fluting. With these stylistic concepts in mind, the following is how temple architecture filters out:
The next lower rank in the taxonomy is Order [after Spaulding 1960]. Order groups temples according to the arrangement of exterior columns.
a. in Antis (with columns only within the naos portico)
b. Naos en Parastasian (with pilasters)
c. Prostyle (with columns in front only)
d. Amphiprostyle (with columns on both front and back only)
e. Peripteral (rectangular colonnaded structure with interior Naos)
Greek temples within established Classes and Orders may further be ranked according to type and subtypes. The number of columns on the temple short side determines the type structure, and a subscript number is assigned to the independent variables, or subtype. This method is outlined below:
Type (grouped according to the number of columns on the short side)
1. Tetrastyle (having four columns on the ends)
2. Pentastyle (having five columns on the ends)
3. Hexastyle (having six columns on the ends)
4. Septastyle (having seven columns on the ends)
5. Octastyle (having eight columns on the ends)
6. Novastyle (having nine columns on the end
Subtype (grouped according to the number of columns on the long side).
The temple of Zeus at Olympia would be classified thus:
The complete classification would then be CLASS, Order, type, and subtype.
I. Class: D …... Doric
II. Order: e …….Peripteral structure (rectangular colonnade with interior Naos).
III. type: 6 …….determined by the number of columns along the width.
IV. subtype: 13 …...determined by the number of columns along the length.
At this point in the taxonomy, a structure either does or does not have Doric-style columns and the associated diagnostic attributes. Subsequently, the new classification establishes specific attribute criteria formalizing a Doric style.
A statistical model was then established employing these reoccurring common attributes and segregating them into discrete data sets. The Order is specific in its determination, and should rule out any structure that might possibly be imprecisely classified as Doric temples. Thus, the Lincoln Memorial in Washington D.C., which has Doric-style columns, is ruled out because it does not have a free-standing peripteral colonnade. The Temple of Zeus at Olympia does have such a colonnade, and is thus assigned to Order e. Because the temple of Zeus at Olympia has 6 columns on the short side, it is grouped in type 6; because it has 13 columns on the long side, it is further subdivided into subtype 13.
Although this classification may seem pedantic, the form is necessary to discern discrete groups of temples. Once those have been established, the ordered data may be subjected to unbiased hypothesis testing through statistical means.