This function is developed in *The Seasoned Schemer* pp. 76-83. It accepts a list `lst` and returns the leftmost atom in it, even if that atom is embedded several levels deep. Any empty lists preceding the leftmost atom are ignored. #lang racket (define (atom? x) (and (not (pair? x)) (not (null? x)))) (define beta (lambda (lst) (let/cc k ; calling k with val will immediately return val from the call to beta (letrec ([aux (lambda (l) (cond [(null? l) '()] [(atom? (car l)) (k (car l))] [else (begin ; each of the following lines will evaluate to '() iff no atom was found in the specified part of l (aux (car l)) (aux (cdr l)))]))]) (aux lst))))) (beta '(((a b) ()) (c (d ())))) ; ~~> 'a (beta '((() (a b) ()) (c (d ())))) ; ~~> 'a (beta '(() (() (a b) ()) (c (d ())))) ; ~~> 'a (beta '(() (() ()))) ; no leftmost atom, returns '() This function could also be written like this: (define leftmost (lambda (l) (cond [(null? l) '()] [(atom? (car l)) (car l)] [else (let ([found (leftmost (car l))]) (cond ; here we check whether the recursive call found an atom in (car l) [(atom? found) found] ; if not, we search for an atom in (cdr l) [else (leftmost (cdr l))]))]))) But in this version, when an atom is found, it is returned back the chain of recursive calls, one by one. The previous version, on the other hand, uses a captured continuation `k` to return the atom immediately upon finding it.