This week a friend told me she was urged to start a survey of a certain practice in period and here is the advice that I offered:
Define
To start, I recommend you sit down and work out the exact definition of the item is for your study. If you are doing embroidery, for example, what would qualify in terms of your survey? Would the stickerie metal work at Birka count? Would a single like of stem stitch across a hem count? Are you doing figurative motifs only or more simple as well? If you are doing tablet weaving, are you only counting metal brocade, or will it be all tablet weaving? Do starting borders from a warp-weighted loom count or are you only collecting information on bands?
Timeframe
Next you need to define your time. Viking Age (by which definition)? Iron Age (which definition/culture)? A solid set of years (example 500-1000CE, or 100-1000CE, or just 9th and 10th century)?
Location (or Culture)
Then determine geography... British Isles, Northern Europe, Eastern Europe, all of Europe? What do you do with pieces that might have been moved across borders. Do you count it based on likely origin or where it was found? If the item you are studying is Saxon, are you looking at those items that were stolen and traded elsewhere, or just those within the British Isles.
Keep Notes
Once you have all that I highly recommend you start an excel sheet to keep a list of each item, making columns for the most relevant things (date of item, culture it originates from, stitch types, ground cloth, fibres used, colors, whatever, size of piece, whatever it is). Make sure you have a column there for the source of your data as well and keep a bibliography with it. Sometimes I also keep a powerpoint for a category where I drop pictures and relevant data (and a note about the source) on each slide. Later I can print these if necessary and shift through them to try to make sense of what I have. (And beyond that, I made notes on the source documents themselves in Mendeley.com)
Context
And then there is the absolutely most important thing when it comes to a survey.... context. How was this thing used historically? Who used it? What was their status? Was it home goods, decoration or clothing? If you want to make resource that others can use as well and nothing is more valuable than that bit of information. An embroidery that was used as part of an ecclesiastical covering does would not be appropriate for lower class clothing. Metal thread tablet weaving is very different in context than a starting border for a piece of cloth. A coin repurposed as a pendant has a different significance than loose coins. Context lets others know whether the item might or might not be applicable to the persona/kit they are building.
Remember that you can always go back and change or add parameters to your survey later. For example, if you set your dates to 9th and 10th century and British Isles only then discover many examples in the 8th century that heavily relate to your initial material, you can expand on your original plan. You can also keep notes of things that might not fit the survey, but that are still relevant in some way and add them as peripheral evidence if that applies.