To investigate factors influencing holistic writing evaluations, writing samples from children in grades two and five were evaluated holistically by a team of teachers. Papers were rated from one, for a poorly written paper, to four, for an excellent paper at that grade level. Correlations were run between holistic ratings and both traditional variables--ideas, form, creativity, mechanics, and wording--and two new variables--expressiveness of language and cohesive harmony. Holistic ratings correlated significantly with student reading and language scores on the Metropolitan Achievement Test, the number of words in the student samples, and the number of errors per 100 words. The use of expressive language, measured by the number of unique modifiers, emerged as a significant variable. No significant correlation, however, was found between holistic ratings and cohesion harmony, indicating that chain interactions were not the best measure of overall writing quality. (MM)