A Brief Review Of Dosage: A Practical Approach
by Steven A. Roman, Ph.D.
Dosage, a technique for classifying Thoroughbred pedigrees by type based on
aptitudinal characteristics inherited from selected sires, originated in the
early part of this century from the research of the Frenchman Lt. Col. J. J.
Vuillier. In his classic study of the extended pedigrees of the best runners
in England and France, Vuillier (in LES CROISEMENTS RATIONNELS DANS LA RACE
PURE) observed that very few stallions appeared with any great frequency.
He called these stallions "chefs-de-race". He also noted that the degree of
inheritance attributed to each of these "chefs" was essentially constant in
all pedigrees, the absolute value (or Dosage figure) varying from sire to
sire. Furthermore, he demonstrated that in successive 15 to 20 year time
frames, new series of "chefs" emerged which eventually established their own
fixed degree of influence. This process, in which new series of "chefs"
periodically become dominant, provides a rational model for the evolution of
the Thoroughbred race horse. Vuillier believed that the objective in breeding
should be to attain Dosage figures in the foal as close as possible to the
established Dosage figures for the breed. For some time, he practiced his
theories successfully in the employ of H. H. The Aga Khan, breeder of such
notables as Bahram, Majideh, Mahmoud and Nasrullah, among many others.
Some years later, the Italian Dr. Franco Varola (in TYPOLOGY OF THE RACE
HORSE) developed a modified version of dosage that retained the principle that
Thoroughbred evolution proceeds through the influence of a very small number
of the stallions at stud in any era. Varola did, however, shift the emphasis
from quantity (i.e., the degree of inheritance associated with individual
sires) to quality (i.e., the pattern of aptitudinal traits inherited from key
ancestors). Discounting the generation in which his expanded list of "chefs"
appeared, he arrived at a distribution of aptitudinal traits in a given
pedigree that described the "type" of the horse being analyzed. A significant
point made by Varola was that the characteristics transmitted by his "chefs"
were not necessarily those they possessed as runners. The focus, instead, was
entirely on the qualities passed on as breeding animals. Thus, in contrast to
conventional pedigree analysis based on a historical perspective of ancestral
performance, Dosage relies on the dynamics of inheritance. As an alternative
and complementary method of pedigree interpretation, it may help avoid
potential problems associated with the traditional concept of "breeding the
best to the best".
Our approach, which first appeared publicly as a series of articles in Leon
Rasmussen's Bloodlines column in Daily Racing Form just prior to the 1981
Kentucky Derby, has been to fuse the basic ideas of Vuillier and Varola,
incorporating both quantitative and qualitative components in the hope of
providing additional insights. In order to establish greater utility, we
chose to use more accessible four-generation pedigrees instead of the
extended pedigrees used previously. We also introduced an approximation of
a genetic effect by halving the influence of any "chef" in each successive
earlier generation. Finally, we established a statistical method for
evaluating the results of our analysis. In this framework, Dosage in its
latest configuration was developed.
"Chefs", chosen on the basis of their observable prepotence for type, are
assigned to one or two of five aptitudinal groups (Brilliant, Intermediate,
Classic, Solid, and Professional) covering the spectrum (from left to right)
of speed to endurance. The assignments are made to best reflect the traits
that these stallions predictably and consistently transmit to their offspring.
For bookkeeping purposes we assign a total potential value of 16 points to
each generation. Since there are, progressively, one, two, four, and eight
sires in the first four generations, "chefs" that appear among these sires
will contribute 16, 8, 4, and 2 points to each as we work back. The points
for all "chefs" present are then tallied in the appropriate aptitude columns.
"Chefs" that confer two aptitudinal characteristics have their points split
between the two aptitudes. In the end, the total points in each column
produce the Dosage Profile (DP), a series of five numbers that reflects the
relative proportions of each of the five aptitudes and is expressed in the
For example, the DP for Sea Hero (Polish Navy-Glowing Tribute, by Graustark) is
3 - 5 - 22 - 4 - 2.
The ratio of points in the speed wing (Brilliant points + Intermediate points
+ one-half the Classic points) to points in the stamina wing (one-half the
Classic points + Solid points + Professional points) is the Dosage Index (DI).
This number is directly proportional to the inherited prepotent speed in a
pedigree and inversely proportional to the stamina. A DI of 1.00 indicates
a perfect balance of the two. The DI of Sea Hero is 1.12 ((3 + 5 + 11)
divided by (11 + 4 + 2)).
If we consider the five aptitude groups as points spaced equally along a
linear scale where Brilliant is assigned a value of +2.00, Intermediate is
+1.00, Classic is 0.00, Solid is -1.00, and Professional is -2.00, the DP
allows for the calculation of the Center of Distribution (CD), that point along
the scale corresponding to the total combined influences of all "chefs" in the
pedigree. In that sense, it is a balance point (analogous to a center of
gravity) of all weighted aptitudes supplied by "chefs" in the four generations.
Calculation of the CD is done by taking the sum of twice the Brilliant points
plus Intermediate points minus Solid points minus twice the Professional
points and dividing that number by the total points in the DP. An exact
balance of speed and stamina yields a CD of 0.00. The CD for Sea Hero is 0.08
(((2 x 3) + 5 - 4 -(2 x 2)) divided by (3 + 5 + 22 + 4 + 2)), which places the
combined effect of all "chefs" in his four generation pedigree between the
Classic and Intermediate aptitudes, although closer to the Classic.
Research using the described methodology as a tool for pedigree classification
has resulted in a number of revealing observations including the following:
There is a direct correlation between the DI or CD and performance at
varying distances as determined from separate populations of stakes
winning sprinters, middle distance runners, and routers. As expected,
the sprinters have the highest values (reflecting the importance of speed
in short races), the routers have the lowest (confirming the need for
endurance in long races), and the middle distance runners fall in between.
Superior Thoroughbreds as a group (e.g., champions, classic winners,
leading sires) have significantly lower DI's and CD's than the general
population of stakes winners, suggesting that outstanding performance
on the track or at stud benefits from a large component of inherited
stamina. There is no evidence of an inherent quality associated with
lower Dosage figures. Rather, the lower Dosage figures reflect the
fact that our most prestigious races are run at longer distances and
that successful competitors are aptitudinally suited to those races.
Only one winner of the Kentucky Derby since 1940 (Strike the Gold),
and only three winners of the Belmont Stakes over the same time frame
(Damascus, Conquistador Cielo, Creme Fraiche) have had a DI above 4.00.
This is in direct contrast to stakes winners in general, of which about
one-quarter to one-third have a DI greater than 4.00 and for which the
average DI is slightly above 4.00. The combination of Dosage and our
discovery that all but four of the winners of the Kentucky Derby since
1972 were ranked as a juvenile within 10 pounds of the highweight on the
Experimental Free Handicap or were named juvenile champion in another
country has become an especially powerful tool in isolating the true
classic contenders. The implications of this result suggest that a
pedigree suited to distance, along with a demonstration of high-class,
early maturity is more important for classic performance than other factors
such as form in the pre-Derby preps at distances less than ten furlongs.
The average DI of juvenile stakes winners steadily decreases throughout
the season, indicating a larger speed component in the pedigrees of
winners early in the year relative to winners later on. This phenomenon
parallels the need for greater stamina as the distances of races for
two-year-olds increase through the year.
The DI of every steeplechase champion since 1972 has been either below
1.50 (consistent with the long distances associated with steeplechase
racing) or above 4.88. This result suggests that the pace of these races
is well within the ability of speed bred runners and that their quickness
over the jumps can be an important factor for success. Surprisingly,
there are no steeplechase champions over this time frame with a DI in the
range of middle distance flat racers.
The average DI of stakes winners at tracks favoring speed is higher than
the average DI of stakes winners at tracks where speed is less favored.
This result is consistent with the observation that brilliantly bred
runners often carry their speed further on the speed oriented surface.
Turf stakes winners have, on average, a lower DI than stakes winners
on dirt. Furthermore, the turf runners have a significantly greater
representation of Solid and Professional "chefs" in their pedigree. This
result is consistent with the observation that turf races are run at a
longer average distance than are races on dirt.
There is an "inflation" factor in the DI and CD over time that parallels
the dramatically increasing influence of Brilliant and Intermediate "chefs"
during the last 50 years, primarily through Phalaris. Accordingly, the
DI's and CD's for successive generations of Thoroughbreds are rising,
although the relative importance of stamina in top class performers
compared to the entire population has remained constant.
The studies discussed here are based on large populations that reflect
statistically significant trends within the total gene pool; therefore, when
applying the methodology to individual horses, we must consider the probability
that the individual will perform according to expectations. A higher degree
of accuracy, as well as an associated predictability, will depend in large part
on a continual modification and refinement of the chef-de-race list, guided by
the principle of a better description of reality.
The Jockey Club Information Systems, Inc. All Rights Reserved