Comparisons With Other BASIC's Speed: TARBELL BASIC will generally run slower than ALTAIR BASIC by a factor up to three, in most tests involving numbers. This is because TARBELL BASIC uses 10 digits of BCD instead of 8 digits of binary. This precludes penny roundoff errors. One place where TARBELL BASIC is faster, however, is in variable and label (line number/descriptor) references. This advantage in speed will not be significant on small benchmark programs, but only on the larger programs, with many variables and labels. The reason for the higher speed in this area is that TARBELL BASIC substitutes pointers for variable and label references, so instead of having to make a lengthy search through a table or program, the item is found immediately by a vectoring method. If you purchase the source, you may notice that several of the subroutines are equivalent to Z-80 instructions. One good way for Z-80 users to drastically improve the speed of their TARBELL BASIC, is by patching in Z-80 instructions for these subroutines. Readability: This is where TARBELL BASIC really shines. Whereas most other BASIC's use line numbers, and are restricted to a few sig- nificant characters in the variable names, TARBELL BASIC allows line descriptors and long variable names. Formatted PRINT output (PRINT USING): Although PRINT USING is not currently part of TARBELL BASIC, it is easier to implement in a subroutine than in most other BASIC's. This is because arguments are allowed for the GOSUB and GOPROC statements, and local variables are allowed by using the PROCEDURE statement. Interpreter vs Compiler: The current implementation of TARBELL BASIC is as an interpreter. This allows the programmer to debug a program online, instead of continually going back and forth between edit, compile, and run operations. It does, however, take up more memory than a compiler. For example, whereas TARBELL BASIC requires about 24k of memory, CBASIC-2 requires only 18k. There is one ray of hope, though. Since we make the source available at low cost, it is quite feasible to remove all those portions of the interpreter that a user doesn't need for a particular situation. Cost: Let's face it, we're way below the competition! L