*maximum value*of a (base ten) positive integer

**N**(with

*non leading zeroes*) such that each of the digits of

**N**, with the exception of the

*first digit*and the

*last digit*, is less than the

*arithmetic mean*of the two neighboring digits.

__Note__: Try to solve this problem analytically, although computer program/ spreadsheet solutions are welcome.