NO LOAD’S TOO HEAVY 
NO CHORE’S TOO TOUGH 
Yes, handle HUNDREDS ol tough, time consuming chores with 
ease! Our perfectly- balanced . all purpose Cart lets you wheel 
BIG. HEAVY loads up to 400 pounds - over rocks, thru mud 
and even UP hills , with JUST ONE HAND! 
That's not all! From hauling away huge loads ol leaves, to 
wheeling trash cans to the curb, this honest to goodness 
WORKSAVER will serve you day in and out, ALL 4 SEASONS 
OF THE YEAR! 
for the rugged, amazingly versatile 
Garden Way Cart! 
The Incredible 
FREE DETAILS. Mail coupon below and we'll send you com 
plete details about all 3 Garden Way Cart Models and show 
you - in actual photos - JUST SOME of the amazing things 
this incredible Cart can do! No obligation, so mail coupon NOW 1 
GARDEN WAY RESEARCH, Dept A847 
Charlotte, Vermont 05445 
I Garden Way Research 
I Dept. A847 
I Charlotte. VT 05445 
YES! Please rush me full, illustrated details about 
I the incredible Garden Carts and include into on low-cost 
I Build-it-Yoursell Cart Kits! 
. Name 
I Address 
I City 
University of California, Berkeley 
RESEARCH 
EXPEDITIONS 
PROGRAM 
1980-81 
• Explore marine environments in 
the South Pacific 
• Study bird behavior off the Cali- 
fornia coast 
• Excavate an 18th C. plantation 
site in Virginia 
• Record Ancient Rock Art 
on Easter Island 
• Study butterfly ecology in 
Australia 
• Observe whale behavior in Baja 
Be an active member of a small 
field research team Your 
participation and tax deductible 
contribution make it possible 
NO PREVIOUS EXPERIENCE 
NECESSARY 
For more Information on theee and other 
expeditions, contact: 
University Research Expeditions Program 
(UREP), Desk NH 
University of California 
Berkeley, CA 94720 or Phone (415)642-6586 
support, Adams again persuaded the 
House to ask Calhoun for an official 
explanation. And again Calhoun wrig- 
gled out, finally delivering a report 
full of obfuscation and rhetoric, and 
still citing the 1840 figures on insanity 
as proof that freedom would be “a 
curse instead of a blessing” for black 
slaves. Jarvis lived until 1884 and as- 
sisted in the censuses of 1850, 1860, 
and 1870. But he never won official 
acknowledgment of the errors he had 
uncovered in the 1 840 census; the fina- 
gled, if not outrightly fraudulent, data 
on insanity among blacks continued 
to be cited as an argument for slavery 
as the Civil War approached. 
There is a world of difference be- 
tween the overcount of insane blacks 
in 1840 and the undercount of poor 
blacks (and other groups) in central 
cities in 1980. First, although the 
source of error in the 1840 census 
has never been determined, we may 
strongly suspect some systematic, per- 
haps conscious manipulation by those 
charged with putting the raw data in 
tabular form. I think we can be rea- 
sonably confident that, with auto- 
mated procedures and more deliberate 
care, the systematic errors of the 1980 
census are at least honest ones. Sec- 
ond, the politics of 1 840 left few chan- 
nels open to critics, and Calhoun’s eva- 
sive stubbornness finally prevailed. 
Today, nearly every census is sub- 
jected to legal scrutiny and challenge. 
Indeed, a few days after Judge Gil- 
more ruled the 1980 census invalid 
until its errors can be corrected, Judge 
Henry F. Werker of Manhattan held 
that the Census Bureau may use sta- 
tistical adjustments (rather than ac- 
tual recounts) to rectify its figures. 
But a strong similarity also connects 
140 years. The prestige of the census, 
invoked with such skill by Calhoun 
to divert his critics, still prevents the 
reform that might curtail its persistent 
and systematic errors. This prestige 
arises from the mystique of quanti- 
fication, the idea that numerical form 
renders information more rigorous, if 
not unassailable. But numbers are only 
as good as the procedures used to 
gather them; if methods are flawed, 
sheer accumulation advances us not 
one whit further toward accuracy. 
Second, the peculiar form of quan- 
tification used by the census gives it 
an air of invincibility. For the Ameri- 
can census is not a statistical operation 
at all. It does not attempt to infer 
properties of the whole from samples. 
It actually tries to count the whole 
by direct, exhaustive enumeration. At 
first glance, what could be fairer and 
less subject to systematic error? Sure- 
ly, a procedure of inference from sam- 
ples offers more opportunity for bias. 
If it were possible (however expen- 
sive) to count the whole with con- 
fidence, then no valid complaint could 
be raised. But it is not, and the very 
attempt to do so engenders a system- 
atic error that guarantees failure. For 
some people are much harder to find 
than others, either by their direct resis- 
tance to being counted (illegal aliens, 
for example) or by the complex of 
unfortunate circumstances that ren- 
ders the poor more anonymous than 
other Americans. 
Regions with a concentration of 
poor people will be systematically un- 
dercounted, and such regions are not 
spread across America at random. 
They are located in the heart of our 
major cities. A census that assesses 
population by direct counting will be 
a source of endless contention so long 
as federal money and representation 
in Congress reach cities as rewards 
for greater numbers. 
Ironically, the common-sense idea 
of simply counting everybody is prob- 
ably the worst, and certainly the most 
expensive, way to conduct a census. 
Schemes of careful sampling and in- 
ference — a perfected version of the 
methods used to compile television- 
rating statistics, for example — would 
do a better job. At least they would 
attempt to deal explicitly with rec- 
ognized biases and might reduce or 
eliminate systematic errors that arise 
from the flawed design of trying to 
count everyone when it cannot be done. 
Censusing has always been conten- 
tious, especially since its historical 
purpose has usually involved taxation 
or conscription. When David, invei- 
gled by Satan himself, had the chutz- 
pah to “number” Israel (I Chronicles, 
chapter 21), the Lord punished him 
by offering some unpleasant alterna- 
tives: three years of famine, three 
months of devastation by enemy 
armies, or three days of pestilence (all 
reducing the population, perhaps to 
countable levels). The legacy of each 
American census will be ten years of 
contention, at least until we realize 
that counting noses is not the best 
way to enumerate vast numbers of 
varied people. 
Stephen Jay Gould teaches biology, 
geology, and the history of science 
at Harvard University. 
24 
