* the entire buffered window, but can't output anything until the size is >=
* 0 (sizes are set to negative while they're pending calculation).
*
- * So SCAN takeks input and buffers tokens and pending calculations, while
+ * So SCAN takes input and buffers tokens and pending calculations, while
* PRINT gobbles up completed calculations and tokens from the buffer. The
* theory is that the two can never get more than 3N tokens apart, because
* once there's "obviously" too much data to fit on a line, in a size