The history of the transmission of ancient Roman texts prior to invention of the printing press is reconstructed from evidence both internal and external to the texts themselves. Internal evidence is used to reconstruct the relationship of the surviving manuscripts of a Roman text to one another, as represented in a modern stemma codicum: a diagram depicting the genealogical relationship of surviving manuscripts and those the stemma's editor believes existed at one time. Stemma are scholar's only road maps to textual connections based on internal evidence, but they may paint a distorted picture of reality because they diagram the relationships of only those manuscripts known or inferred today. If surviving copies are few, the stemma perforce brings into proximity manuscripts that were widely separated in time and place of origin. Conversely, the stemma can also bestow a semblance of separation on manuscripts written within a few months of one another or even in the same room. One type of external evidence that may shed light on the transmission of Roman texts is the availability of a work in the Middle Ages, when many classical texts were circulated. Too often, though, too much is inferred about a particular work's circulation in the Middle Ages from the number of manuscripts surviving today. When a work survives in a single manuscript copy, editors call the manuscript, rather glamorously, the "lone survivor" – implying that all its (presumably rare) companions were destroyed sometime early in the Middle Ages by pillaging barbarians. It is equally possible that the work survived far into the Middle Ages in numerous copies in monastic libraries but were unnoticed due to lack of interest. The number of extant manuscripts, however few, really does not allow scholars to infer how many ancient Latin manuscripts of a work survived to the ninth, the twelfth, or even the fifteenth century. Quotations from a Roman text by a medieval author are another category of external evidence: but does the appearance of a rare word or grammatical construction – or even a short passage – really indicate a medieval author's firsthand knowledge of this or that ancient work, or does such usage instead derive from some intermediate source, such as a grammar book or a popular style manual? Medieval authors do quote extensively from ancient authors; while such quotations provide some evidence of the work's medieval circulation, as well as define its evolving fortunes and the various uses to which it was put, they may be far less useful in reconstructing the text of an ancient work. Much as scholars want to look for overall patterns and formulate useful generalizations, the transmission of each text is a different story and each manuscripts history is unique. Scholars must be careful not to draw conclusions that go beyond what the evidence can support.