and finance. After 622 A.D., competitive tests were administered on three levels.
Candidates first had to pass a local district assessment before they could compete at
the provincial capital. Those who passed at the provincial level were honored with
a twenty-four hour exam in Peking, after which they went home to await the official
report on their place in the "scale of merit."

Assessment changed little in the following centuries until the rise of a technocratic
civilization. In America, until about 1850, most examinations were limited to
recitation and long, difficult essays ( Davis, 1971; Lien, 1976). The lack of reliability,
implicit in scores from these assessments, was apparent as early as 1845 when
Horace Mann argued in favor of the "new type examinations" because of their
objectivity ( Ruch, 1929). However, it remained for researchers in the early 20th
century to empirically support Mann's point ( Buch, 1916).

Accountability and standardized forms of evaluation quickly became issues in
measurement and evaluation. Mann, who was to become a major influence in the
development of assessment, ( Davis, 1971; Ruch, 1929) wanted to develop an
instrument that would measure the progress of pupils in the Boston school system.
Principals were to be fired if their students did not show satisfactory results on the
tests. The wave of immigration into the United States during the mid-19th century
exerted a subtle pressure on the development of assessment. There was a desire to
sort out the "rough," uneducated new immigrants by using "scientific"
measurement instruments.

Many educators argued for the development and measurement of educational
standards. ( Boten, 1932: Buch, 1916; Monroe De Voss & Kelley, 1917). Tests were
developed for spelling, language and arithmetic skills ( Boyington, 1932;
Buckingham, 1916; Courtis, 1913, 1916a, 1916b, Gray, 1923; Hall, 1911; Monroe,
1923; Pryor, 1923). The "new type examinations" estimated a student's knowledge
by the number of words spelled correctly or the number of problems completed
successfully in a given period ( Ebel, 1970; Thorndike, 1911, 1912, 1918;
Washburne, 1922).

Psychology at this time was developing into a "rational science," and
mathematics as well as statistics began to invade what had been thought of as a
purely subjective realm. Inevitably, calls for quantitative and content standards were
voiced by well renowned educators ( Thorndike, 1911, 1912, 1918; Washburne,
1922). Around the turn of the century, two milestones in modern assessment
occurred. In the late 1800's, Francis Galton established his laboratory in South
Kensington, England. Galton measured and recorded individual differences in
human physical characteristics and sensory and psychomotor responses. His results
seemed surprising when published in 1869 and 1883. Individual differences were
distributed according to a mathematical model called the "normal curve of error"
( Harris, 1966; Hashway, 1977c; Freund, 1962). This discovery led many to believe
that mental and physical attributes must also follow the "normal curve." If the
results of an assessment were not normally distributed, they inferred that there was
something wrong with the test-as opposed to inferring that whatever the assessment
was measuring was not normally distributed. Ruch ( 1929) attributes the normal

-2-