16th February 2014, 04:22 PM
(This post was last modified: 16th February 2014, 04:23 PM by barkingdigger.)
Wow Kevin, what an interesting can of worms! For a start, this could be split into two almost independent issues - single context "concept" and "practice". And frankly most of the heated discussion I've heard centres more on practice than concept.
As a concept, giving each feature a unique ID allows us to relate them to other features temporally and spatially, as well as relating them by activity in order to build up the story of the site. This is the essence of grouping and phasing that allows us to set out a chronology that can be tied to dating evidence and can be independently checked by others. Even when we dig in "spits", or arbitrary layers, we ultimately want to know how features relate to one another. There are of course limitations that may force us to dig in spits or box-sections etc. But it is hard to see how choosing to dig in spits can offer as much relationship info as digging one context at a time - therefore to my mind it is always a second-class option, to be taken when "plan A" isn't possible. Of course, one of the drivers for recording philosophy can be cash - single-context digging/recording is labour-intensive and requires a wide range of skills among the digging staff.
Practice is where things get ugly within the single-context world. Many folk have a hard time seeing the difference between single-context "understanding" and paper record-creation. Some systems use the context number as the plan and photo numbers, effectively forcing diggers to draw only one feature per sheet regardless of how silly it looks or how much permatrace it wastes. (Been forced to record a floor-full of obviously related stake holes one per sheet in the past - dumb idea, but their choice...) This may make it easier to create a matrix by stacking plans without the need for a drawings index, but it seems overkill to me. And while it is ideal to dig features in strict chronological reverse-order, it is rarely practical. The issue is easily alleviated by using separate IDs for all records, and then listing which contexts appear in them, but this does come with a burden in terms of extra effort. And there will be times when the sequence breaks down, through truncation or "missed" observations, but a flexible brain can sort these out.
New technologies such as databases, GIS, etc rely on strict adherence to rules, which favours single-context systems. Yes, it can be a pain to have to identify the whole outline of a feature, and it may not even be possible under certain conditions, but a good system will have tools for dealing with these circumstances as they arise. ("edge uncertain" or "truncation" lines spring to mind in paper drawings...) I have no problem with multiple contexts occupying the same permatrace as long as they don't obscure one another, but I still want them to be individually ID'd and put into the matrix if possible.
I see no reason why we need to keep up a pretence that there is an "either/or" ideological choice here - surely single-context is the ideal for obvious reasons of clear understanding, but spits or other techniques can be the best (or only) way to go on some sites.
Your mileage may vary...
As a concept, giving each feature a unique ID allows us to relate them to other features temporally and spatially, as well as relating them by activity in order to build up the story of the site. This is the essence of grouping and phasing that allows us to set out a chronology that can be tied to dating evidence and can be independently checked by others. Even when we dig in "spits", or arbitrary layers, we ultimately want to know how features relate to one another. There are of course limitations that may force us to dig in spits or box-sections etc. But it is hard to see how choosing to dig in spits can offer as much relationship info as digging one context at a time - therefore to my mind it is always a second-class option, to be taken when "plan A" isn't possible. Of course, one of the drivers for recording philosophy can be cash - single-context digging/recording is labour-intensive and requires a wide range of skills among the digging staff.
Practice is where things get ugly within the single-context world. Many folk have a hard time seeing the difference between single-context "understanding" and paper record-creation. Some systems use the context number as the plan and photo numbers, effectively forcing diggers to draw only one feature per sheet regardless of how silly it looks or how much permatrace it wastes. (Been forced to record a floor-full of obviously related stake holes one per sheet in the past - dumb idea, but their choice...) This may make it easier to create a matrix by stacking plans without the need for a drawings index, but it seems overkill to me. And while it is ideal to dig features in strict chronological reverse-order, it is rarely practical. The issue is easily alleviated by using separate IDs for all records, and then listing which contexts appear in them, but this does come with a burden in terms of extra effort. And there will be times when the sequence breaks down, through truncation or "missed" observations, but a flexible brain can sort these out.
New technologies such as databases, GIS, etc rely on strict adherence to rules, which favours single-context systems. Yes, it can be a pain to have to identify the whole outline of a feature, and it may not even be possible under certain conditions, but a good system will have tools for dealing with these circumstances as they arise. ("edge uncertain" or "truncation" lines spring to mind in paper drawings...) I have no problem with multiple contexts occupying the same permatrace as long as they don't obscure one another, but I still want them to be individually ID'd and put into the matrix if possible.
I see no reason why we need to keep up a pretence that there is an "either/or" ideological choice here - surely single-context is the ideal for obvious reasons of clear understanding, but spits or other techniques can be the best (or only) way to go on some sites.
Your mileage may vary...