Abstract

A palynological data storage and retrieval system is briefly described. Iy is currently being applied to 900 records of Formosan pollen grains. First, procedures which were used for encoding the morphological features of a pollen grain are outlined. Next, a storage and retrieval system is discussed, which organizes and updates reference pollen records, catalogs them, and enables retrieval of those stored records which share the same features with an unknown specimen. The whole system has been proved useful and effective. The ways in which the variable characters are treated for the sake of obviating encoding errors are highlighted.

中文摘要

花粉具有萌芽口、雕紋、形狀等特徵以及不易摧毀之外壁，致使其成研究植物分類學、古植物地理學、空中孢粉學、考古學與地層學所不可缺之一環。由於花粉特徵之變異性大，同時許多不同種類能產生類似之花粉型，加上經常受到觀察角度之限制，使得花粉之鑑別困難重重。因此本論文之目的即在尋求一迅速有效之方法，以辨識花粉之種類。全文分為兩部分。第一部分將易於分辨、且具有辨別價值之特徵及特徵類別盡量予以蒐集。然後有系統地加以歸類編碼。歸類後之花粉特徵可歸納為主、副兩辨識特徵。前者依其變異性之大小復分為固定、變異兩類 ; 後者依其性質劃分為甲、乙、丙、丁四羣。所有特徵計31項，而特徵類別則超過168種。此種歸類方式，不僅適應花粉之特性，同時亦符合查詢系統設立之需求。第二部分在設立一套查詢辨別系統。整個系統由五項目所組成 : (1)建立適合快逐查詢之花粉檔案。此檔案包括資料、索引及串列三部門 ; (2)新花粉紀錄之加填 ; (3)錯誤紀錄之修正 ; (4)錯誤紀錄之刪除 ; (5)未知花粉之查詢。