ON ERRATA RECEPTA. 145 



not pictorial, but an arrangement of points shewing the numbers to be 

 named. Somewhat thus : 



These groups of points rapidly made, and each respectively con- 

 nected together by a tracing of the calamus as it passed quickly from 

 one dot to the next, may be conceived of as developing at last into our 

 present Arabic numerals, the line connecting the points denoting also 

 perhaps the order which the eye of the enumerator would swiftly^ 

 follow. 



This line itself may have been suggested by the accidental marks 

 left by readers in the act of calculation. The so-called nailed letters 

 in inscriptions are formed by light straight lines connecting bold punc- 

 tures which mark out the general form of each character. This pro- 

 cess of course produces a set of letters that are angular. In an in- 

 teresting alphabet of the time of the Seleucidse (about B.C. 250) the 

 characters are marked out by an increased number of dots, with light 

 lines connecting them, forming the letters called perJees by the French, 

 from their headed appearance. In these the angles are converted into 

 curves in such letters as B and O. In a similar manner the numerals 

 formed from the dots of computation speedily had their angles con- 

 verted into curves, approximating thus to the flowing forms of our 

 present cyphers ; just as in rapid writing, the angular capitals also 

 become at length the so-called round hand or cursive script. 



The symbol for seven, about which on this hypothesis a difficulty 

 may present itself, is either a combination of the written 6, with a 

 connected point below for plus one ; or an adaptation of the Greek seta 

 which, though standing sixth in the present Greek alphabet, is in no- 

 tation the symbol for 7, one letter, bau, i.e. the digamma or/, having 

 been disused as a letter, though retained as a symbol for 6. It will 

 be noticed also that the final cypher has the value of ten, which may 

 help to render rational the notation 10, 20, 30, &c. 



I am aware of the theory that the original elements of the Arabic 

 numerals were strokes or tallies, corresponding in number with the 

 quantities indicated, productive also, in the first instance, of a set of 

 square or angular characters. As their origination in points was inde- 

 pendently conceived, and is at least equally probable, the supposed 

 process has been briefly detailed. It may here be added that al- 



