A CADD model is normally divided into some number of segments, each of which can be turned on or off for viewing and editing at any time. These data segments are crucial to the effective use of CADD, since they make it possible to view only the portion of a model which is desired at a specific moment. Using the names of the data segments, one may turn on, for instance, the portions of the model which are related to a specific time period, the portions containing specified artifact types, the portions which comprise a specific part of the structure,the portions which are made of particular building materials, etc. These data segments are called layers in CADD systems, but they have no relationship to excavation strata; they are simply groupings of parts of the model.
Regular readers of the Newsletter have already seen considerable discussion of the importance of effective layer names in CADD systems. If the layer names are properly defined, it is easy to make fine distinctions within the model, for instance, to separate blocks of marble from blocks of limestone or chipped stone artifacts from ceramic ones. Moreover, a carefully crafted system will permit the operator to utilize these distinctions efficiently and easily when working with the model. Selecting layers which meet specified criteria can be very easy if the system is well designed, very taxing if not.
The CSA Layer Naming Convention has been published previously in the Newsletter; it defines a system which is meant to provide both the appropriate segmentation of the model for analytic use and the necessary ease of use.
The key to the utility of any layer naming convention is matching the naming convention to both the needs of the user and the searching capacity of the computer system. That is, if the system is to search for layers according to analytic needs, it can only do so if the names of the layers reflect the capabilities of the computer system and the analytic needs of the user. Fortunately, the capabilities of the computer system are very clear. One may request that layers be turned on or off by naming them explicitly -- by typing the layer names from the keyboard -- or by using special characters, often called wild-card characters, which the computer will use to select groups of layers instead of individual ones. Specifically, computer systems understand that a question mark may be a stand-in for any one, unspecified character, but only one character. Similarly an asterisk may be a stand-in for any number of unspecified characters (including none).
So, consider this simple system of layer names for architectural remains at a site (a full layer naming system
would be far more complex). Each layer name consists of five characters, and each character's name is derived from
the following choices:
Character 1 - T for temple, P for propylon, S for stoa
Character 2 - W for wall, C for column, S for step
Character 3 - M for marble, L for limestone, W for wood
Character 4 - A for phase 1, B for phase 2, C for phase 3
Character 5 - A for in situ, B for scattered, C for re-used elsewhere (The same blocks could be modeled on two layers -- on one layer in their final positions and on another layer in the positions of prior use.)
The position of the character and the character itself combine to give meaning. An S as the first character means the material on that layer is from a stoa; an S as the second character means the items on that layer are steps. Similarly, an A as the fourth character means building phase 1; an A as the fifth character means in situ material. Using position and character together makes it possible to make many more distinctions from one to another layer; it also makes the searching more efficient.
Now consider some layers from a hypothetical model (a model would have many more layers.):
Layer #1 - TWLAB
Layer #2 - TCMCA
Layer #3 - TWMBA
Layer #4 - TSMCB
Layer #5 - TCMBB
Layer #6 - TSMBA
Layer #7 - TSLBA
Layer #8 - TSLAC
Layer #9 - TWMCA
Layer #10 - TCWAC
Layer #1 consists of temple wall remains made of limestone, belonging to the first building phase, and found scattered around the site. Layer #2 consists of temple column(s) of marble, from building phase 3, and found in situ.
All the layers have temple remains; some have wall blocks (#1, #3, & #9); others have column drums (#2, #5, & #10), and four have step blocks (#4, #6, #7, & #8). Only layer #10 has material of wood; layers #1, 7, & 8 have material of limestone; the other have material of marble. Some are from phase one (#1, #8, & #10) and some from phase two (#3, #5, #6, & #7). Three are from phase 3 (#2, #4, & #9). Some remains are in situ -- layers #2, #4, #6, #7, #9; some scattered -- #1, #4, and #5; the rest are in positions of secondary use - #8 & #10.
Putting the layer names to use involves selecting layers of interest in an efficient manner so that the computer system can do something with the group selected. For instance, to ask the system to turn off all layers with wall blocks, one must understand that the second character shows whether the layer deals with wall blocks, column drums, or step blocks. Thus, one would construct a pattern specifying W (wall) for the second character, the only relevant one, to specify all layers of interest -- without the user having to know how many there may be or what their full names are. The pattern ?W??? (or the pattern ?W*) would instruct the system to search for and turn off all layers which have a W in the second position. (The pattern can have a question mark for each unspecified character or an asterisk for any number of unspecified characters.) Layers #1, # 3, and #9 would be turned off. To select the layers of limestone, one would use the pattern, ??L*, since only the L in the third position matters; the system would select layers #1, #7, & #8. A search for the pattern ?S* would mean all layers with step blocks would be selected - layers #4, #6, #7, & #8. Of course, one could -- and usually would -- combine requirements to ask the system, for instance, to select all layers with limestone blocks found in a position of secondary use. The search pattern would be ??L?C, and layer #8 would be selected.
Since even a simple model may contain a hundred layers, the use of such a system is crucial to manipulating the model. That is why the CSA Layer Naming Convention was developed and why it has often been discussed in the Newsletter. It was intended to be inclusive, but it was also made flexible so that others could use the core of the system without being constrained by its details.
The original CSA convention was based on the searching capabilities of DOS, the basic software for most IBM personal computers and their work-alikes, and DOS does not permit maximum flexibility, because it will ignore any character specification after an asterisk in the pattern. That is, a search for the pattern *A, using DOS, would be the same as just an asterisk, and it would turn up all the layers instead of all the layers ending in A -- all the layers with in situ material. (Sophisticated users of DOS will note the slight exaggeration. DOS will recognize a period as a break in the pattern.)
Starting with Release 11, AutoCAD® makes it possible to use an asterisk anywhere in the search pattern. Thus, the search for *A, using AutoCAD, indicates a group of layers ending in A; AutoCAD would correctly identify layers #2, #3, #6, #7, & #9. Similarly a search for ?C*A would locate only layer #2, the only layer with in situ remains of column drums.
This use of the asterisk means that a layer name can be far more complex yet still permit easy searches. In fact,
two different layer naming conventions can be combined. This can be achieved by using a character which is not
otherwise used in the naming convention as a marker to separate two sets of names, each of which may be based on
a different naming scheme. Thus, using the underscore (_) as a marker, we could graft two schemes onto one another
and search for only one, taking advantage of the expanded use of the asterisk. Consider the layer names below.
To the names used above we have added an underscore and a simplified version of the kind of system a site
architect might use during the excavation and before the analysis has begun - a single letter for the season of
excavation (A for the first season, B for the second, and so on), a letter designating the trench, and another
letter representing the name of the excavator in charge of the excavation of the material on the layer.
Layer #1 - TWLAB_AAE
Layer #2 - TCMCA_ABF
Layer #3 - TWMBA_BCF
Layer #4 - TSMCB_ADE
Layer #5 - TCMBB_BEF
Layer #6 - TSMBA_AFE
Layer #7 - TSLBA_AGF
Layer #8 - TSLAC_CHE
Layer #9 - TWMCA_AJF
Layer #10 - TCWAC_CKF
Should one wish to use the scheme for the analytic layer names (prior to the underscore), one could search by using the patterns already shown above, simply making sure that each pattern ends with an asterisk so that the underscore and following letters are ignored. However, if one wished to use excavation season, trench, and/or excavator as the search criteria, one could make the system by-pass the first six characters by starting each pattern with *_. A search for *_A?, for instance, would turn up layers #1, #2, #4, #6, A#7, and #9 - all the layers excavated in the first season. A search for *_BE? would turn up layer #5, the only layer with material from the second season and trench E.
Why does anyone need such complexity? The advantage is that the system permits a site architect to maintain one convention while the excavation is in progress, to add another for analytic purposes, and then to use both - to maintain two conventions indefinitely. Using a double convention permits the transition from excavation to analysis to be easier; it also permits searching by either convention. If both conventions are maintained indefinitely, users familiar with the excavation convention would not be encumbered by the need to learn the analytic one; nor would they lose the ability to access layers by criteria having to do with the manner or time of excavation.
The value of this system was discovered while defining the naming scheme to be used for the Gordion drawings. Phoebe Sheftel, Director of the CSA Gordion Project, wanted to use the CSA Layer Naming Convention (as if she had a choice), but she also wanted to maintain references to specific paper drawings and, where possible, to specific buildings. So the Gordion convention will have the standard CSA scheme (modified to use periods rather than dates, since dates have not been finally determined), then an underscore, then a drawing number, then an underscore, then a building name. Three different conventions will exist together, and users will be able to search using any one -- or all three. To use the standard CSA Layer Naming Convention, one would use an appropriate pattern and finish with an asterisk. To refer to a drawing number, one would start with *_ and add the drawing number and another asterisk; to refer to a building name, one would start with *_*_ and add the building designation.
This more complex way of naming layers will reflect more fully the complexity of archaeological finds, but it adds relatively little complexity to the use of the system.
The new CSA Layer Naming Convention, with the addition of this new double- or triple-name possibility is available from CSA on request.
For other Newsletter articles concerning the applications of CAD in archaeology and architectural history or the "CSA CAD Layer Naming Convention," consult the Subject index.
Table of Contents for the Feb, 1993 issue of the CSA Newsletter (Vol. 5, no. 4)
Table of Contents for all CSA Newsletter issues on the Web
|CSA Home Page|