We have put up at the UPWARD headquarters in Los Banos, Laguna, a Memory Bank filing cabinet. Under separate color coded folders (by place of collection) for each locally named variety, we have placed the technical characterization (Huaman 1991) and scientific illustrations of the sweet potato variety and its parts (done by a trained taxonomic illustrator), side by side with the farmer characterization evaluations and drawings. Other information available on traditional sweet potato varieties includes two files on indigenous beliefs and practices associated with sweet potato, several files on local evaluation criteria for existing and recall varieties, a modest archive of taped and transcribed life histories of elderly male a female farmers, and agricultural calendars of cropping systems of subsistence and commercial oriented farmers in Bukidnon.
There is still a long way to go in terms of both theoretical and applied aspects of memory banking. First, there is a need for verificatory studies on the correspondence between varieties and names. How much of the terminological confusion can be accounted for by different informants giving a particular variety different names? How much of it is due to assigning one name for what are actually several distinct varieties? Does a one-to-one correspondence exist between local names and the samples in the germplasm collection? We have systematically collected actual specimens and documented the different varieties cited by informants. We have preserved these herbarium "types" as well as living specimens in the local germplasm collection. Now, there is always a concrete specimen that can be used to help clarify most redundancies and certain taxonomic disagreements.
Another area that is worth investigating further both for its application in agricultural research and development and its theoretical implications in anthropology is the issue of cultural variation in cognition (e.g., classifications, evaluations, and beliefs) and performance (e.g., technologies and practices). We observed an interesting complexity and diversity in local perceptions with respect to sweet potato and its cultivation. There was a multiplicity of descriptions and criteria regarding local varieties as well as occasional absence of consensus on "appropriate technologies." Is this a case of idiosyncratic individual variation or of patterned variation? If it is patterned, what is (are) the significant determinant(s) of the patterning? Is it primarily ecological (including geographical location and resource endowment) or is it more socioeconomic (including demographic and status parameters)? If the variation is purely idiosyncratic, then it can be largely ignored. But if a significant degree of patterning underlies the observed variation, then it is a clear argument for intracultural variability and a call for "tailoring" agricultural recommendations on varieties and technologies to particular user groups.
Apart from analysis, some challenges in research design still have to be addressed. One is how the potential for comparability and generalization using informal techniques can be increased. It is recognized that the effect of researchers' personalities and interviewing style on their interaction with informants and, as a result, on the quality of information elicited cannot be eliminated altogether. Nonetheless, for the memory banking techniques to be truly useful in eliciting comparable information, they should be made "researcher-proof to the greatest extent possible. In other words, any other researcher with adequate training and commitment should be able to replicate the findings. A lot of work still has to be done in data storage and retrieval in consultation with experts in information systems to streamline the system and increase its efficiency. In this direction, the use of relational data base to allow simultaneous access to information in different files is an exciting possibility.
Finally, for germplasm collections to be of greater benefit to the actual users, the collections need to be moved closer to the source. The local germplasm collection Dalwangan, Malaybalay, is a step in that direction. But the process needs to be completed by initiating in situ germplasm collections run with farmer groups, mainly native populations, that have propagated and maintained these varieties for a long time. The same is true for knowledge systems. Ideally, the present generation should assume the role of custodians of information and effective transmitters to the next generation, with scientists playing only a supportive or facilitative role.