The
Precision,
Accuracy,
of Automatic
automatic
continuous-flow
six constituents,
Two parameters
mental
drift.
UTOMATIC
ing
increased
chemistry.
lyzed
ous”
were
Oglesby
was evaluated
influenced
technics
ANALYZERS
In
widely
many
Methods
under
routine
daily
use for
is shown
by the
available
to improve
amount
of
the accuracy
sample.
of the auto-
system.
A
been
when
also
of correction
matic-analysis
system
K. M.
Errors
sodium,
potassium,
chloride,
carbon dioxide,
urea, and glucose.
have been quantified,
interaction
between
samples and instruResults
Application
and
Inherent
Flow
Continuous
R. E. Thiers
An
and
use
in
the
latter
based
on
analytical
accepted
different
field,
one
for routine
samples
for specific
rather
than
constituents
a “batch”
continuous-flow
chemistry
methods
and
the
instrument,
gain-
clinical
has
in
AutoAnalyzer,*
use.
These
instruments
of similar
composition
because
basis
and
are
particularly
they
operate
can analyze
are
must
on
many
valuable
be ana-
a “continusamples
in
sequence.
Certain
propei-ties
ferent
from
those
to these
Particularly
of
familiar
pi-opem-ties
in the
continuous-flow
to the average
can lead to serious
loss
field of clinical
chemistry,
practice
to make
ent, mishandling
only
one determination
of an automatic
instrument
good
are,
ler
From
data
these
the
which
collditions,
Clinical
in fact,
ca
Chemistry
produce
Laboratory,
significantly
systems
analyst.
are
inherently
difLack
of attention
of precision
in which
per
specimen
can provide
in error.
incorrect
data
Department
of
and accuracy.
it is common
very
Biochemistry,
per
constituapparently
Automation,
prolifically.
Duke
Medical
Center,
Durham,
N. C.
Presented
before
the
14th
Annual
Meeting
of
the
American
Association
Chemists,
Aug.
27, 1962,
Santa
Monica,
Calif.
Thanks
are due to many
members
of the Clinical
Chemistry
Laboratory
whose
increased
by this
study,
particularly
to Mrs.
Gerry
Naylor.
Received
for publication
Oct.
5, 1962.
AutoAnalyzer,
Technieon
Corp.,
Chauncey,
N. Y.
246
un-
University
of
Clinical
work
was
Vol.
ID,
In
ied
No.
3,
this
the
conditions
of
Two
quantified;
addition,
of AutoAnalyzers
routine
parameters
operation.
sample
effects,
and
Technics
for
routinely
for
be
correcting
them,
carbon
dioxide
were
and
of
performed
of modules
for
accuracy,
have
by
errors
have
due
been
been
to these
developed.
shown
routine
111
been
drift.
In
tile amount-
to improve
use.
and Apparatus
Standard
AutoAnalyzer
modules
out this
work.
The
determinations
combination
systems
instrumental
the
for
stud-
for six determiand
carbon
di-
influenced
measuring
of these
correction
technics
have
an(l accuracy
of the utoAlIalyzer
Reagents
and
and
been
and
evaluated
chloride,
continuous
between
samples
that
results
may
available.
Applications
the plecisloll
to
has
Precision
have
been
potassium,
peculiar
interaction
it -as
found
247
ERRORS
performance
recommended
conditions,
urea,
glucose,
sodium,
oxide.
of
INHERENT
research
under
under
nations;
1964
reagents
sodium,
on
these
an
four
were
used
potassium,
“Electrolyte
throughchloride,
Analyzer,”
simultaneous
(1).
The
urea
and
each sample
by the
heating
baths
were
glucose
were
also
determined
simultaneously
recommended
method
(2) except
that separate
employed,
and a 6-mm.
flow cuvet
was
used
urea
instead
of the
recommended
were
utilized
exactly
as
8-mm.
recommended
Experiments
This
of
study
these
vised
such
standard
of
(see
Amount
of Sample
cup
manufacturer.
of the
of
used
part
of the
separate
instrument
routine
analyses.
of daily
routine
quality
routine
use
were
as were
of the
records
as
daily
experiments
de-
observed
Precision
and
blind
analyses
control.
“drift
The
control
acof
same
stand-
Available
which
the amount
upon
the
is present
as
of
results
from
were
the
amid methods
below).
AutoAnalyzers,
plastic
wholly
number
properties
samples
by
Reaents
on
95#{176}
for
and Discussion
samples-part
serum
ards”
that
effect
out
A
an effect
upon
the
data
were
obtained
serum
In
carried
instruments.
to measure
to have
curacy
type
was
cuvet.
a
determinations
for
can
of
final
the
the
sample
is
hold
slightly
more
than
in
the cup
of course
sample
result,
analysis
placed
provided
(3-6).
placed
in
a
2 ml.
has
that
small,
disposable,
It has
been
stated
only
a negligible
sufficient
sample
248
THIERS
An
experiment
result
upon
parameter
rather
than
sufficient
pared
fronl
was
the
performed
amouiit
for
the
0.8, 0.8, 0.5,
shown,
and
tiolled
above.
were
chart
0.5, 0.3,
and
determinations
Tile results
the
for
Staildar(lS
the first
effect
peaks
of
milliequivalerit
0.3
ml.
were
obtained
of
the
directly
made
are
strip
the
or
Q3
104
4
34
cups
was
143
#{149}..
obtained
were
100
0,5
0,8
1.0
145
146
149
1.5 2.0
r
r-,,
to
The
reading
the
from
results
of the
nearest
figures
i.
I5I usrEq/L.
-:
SOOIUM
#{149}-i-----
f---
I
II
II
II
II
11- 54-H-
II
t__
II
II
II
III
IIIII Ill
-
r”
-
ti
,
Il 111111
II
II
II
II
fff1lff1llf1Jl
II
+1+11
II 44414
-+
L
Fig.
strip-chart
1.
Effect
record.
of amount
of sample
rnniiiiiii
in AutoAnalyzer
iiii i
cup
on result
of
sodium
the
for
by
obtained
read
ml.
144
,..,
pre-
were
aliquots
1.5, 1.5, 1.0, 1.0,
curve
per
the
in the cup
more
than
obtained
data
Results
the
that
analysis
for that
day.
The
series
were
not used
because
milligram
2
RESULT
of
recording
tile
of
likely
iun
in tile order
six constituents
Inenin Figs.
1 and 2. Fig-
calibration
below).
liter
L.
chart
2 shows
from
(see
samples
of the
shown
Chemistry
dependence
seems
A series
Tile
used
in the routille
two samnples
of each
per
(it
volumes,
all of which
serum
; 2.0, 2.0, 2.0, 2.0,
Figure
interaction
the
cup.
present.)
the following
of standard
ure 1 is a reproduction
sodium
deterniination.
tile
evaluate
in the
is actually
the depth
of sample
since
in all hut two cases
amounts
analysis
colltaining
a single
pool
to
of sample
being
tested
the amount,
Clinical
& OGLESBY
determination;
show
Vol.
10,
the
No.
results
plotted
as
Figure
depth
the
3,
1964
INHERENT
of
at
least
four
averages
amid
2 shows
clearly
on
results.
result.
it
The
is
249
ERRORS
replicate
experiments
at
that
gi-eater
there
the
is
depth
a
definite
noteworthy
that
the
tile
0-02-
,ia
n.E
SODIUM
+5
0
0I
I-.
o
-5-
-tO-
SAMPLE
Fig.
2.
of
sample
effect
of sample,
+02-
o,
point,
extremes.
particularly
_j
each
Effect
of
VOL..
SAMPLE
VOL.,
amount
of
in ml.
in ml.
sample
on
data.
higher
error
will
from
be
this
250
THIERS
effect
can
be
theoretically
ple
cup.
picked
quite
large
instrument
0.5 ml. in the cup, very
of the effect
of sample
in the
even
when
sufficient
for the
For
the
electrolyte
up by the
cup
after
the
& OGLESBY
of the
It is not
limited.
for each
cup,
the
In such
analysis.
rum
is utilized
ta in ed.
has
results
removed
but,
even
still
in clinical
occasion
show
required
with
are
0.4
that
practice
theoretical
or
volume
required.
employed
for
of
to
the
the
error.
the
sample
low
results
volume
will
to place
roughly
to use amounts
When
standards
When
to make
from
is
to use minimal
volumes
minimal
volume
of se-
seriously
of the nominal
used,
the volumes
aliquot
the tip
possible
a significant
chemistry
demands,
be identical.
be utilized
its
0.5 ml., and although
close
as is physically
precision
it is essential
in the sample
cups
and
should
2 may
which
present
in the samthe
sample
volume
0.3 ml.
For maximum
umes
of samples
cal and
Figure
sample
Chemistry
significant
negative
errors
can occur
because
depth.
Although
0.2 ml. of sample
may be left
a case it is comnon
Clearly,
if the
as
of
are
is nominally
instrument
infrequent
amounts
determination
determinations,
a cup which
originally
contained
sampling
capillai-y
is placed
as
bottom
Clinical
be
ob-
similar
volwell in excess
smaller
amounts
must
be
and samples
become
criti-
necessary,
corrections
curves
such
for variations
as
those
of
in sample
volume.
Interaction
Samples
analyzed
on the
AutoAnalyzer
teract
with
each
other
in such
rect data.
This
effect
is perhaps
for urea,
in which
determinations
of
serum
demand
Figure
from
uremic
patients
are
not
independent
a way as to produce
most
often
recognized
which
happen
to
may
be
so
but
in-
occasional
incorin the method
follow
specimens
obviously
in
error
as
to
repetition.
3 illustrates
this
phenomenon
sium.
The
elevation
of the result
for
produced
by the preceding
5.4 mEq./L.
Experiments
action
between
in the
determination
of potas-
the second
3.1-mEq./L.
sample
is very marked.
have
been
performed
which
samples
may
be expressed
demonstrate
quantitatively
sample
that
in
intersimple
fashion
and that
this information
may
be used
to correct
the errors
that result
from
interaction
between
samples.
Samples
containing
known
amounts
of urea
and
of glucose
were
run,
preceded
by known
samples
containing
various
concentrations
of
these
constituents,
from
very
low
to very
high.
The
difference
be-
vol.
10,
No.
3,
tween
the
preceding
4 shows
ceding
1964
INHERENT
known
sample,
these
sample
values
‘as
and
taken
the
as
251
ERRORS
results
the
errors
as a function
and of the measured
after
following
the
to interaction.
Figure
obtained,
error
of
due
concentration,
sample,
for
both
urea
of
the
nitrogen.
pre-
These
Potossium
36
mEg/I
3.1 mEg/I
mEg/I
3.1
H20
H20
Fig.
3.
between
Interaction
samples.
E
820
Concentration
8
0-100
of fpiping
Samples
mg/IOOmI
6
8
4
2
IO
02040
Fig.
the
data
error
4.
6080
tOO
CONCENTRATION
Direct additivity
the method at
of
imply
that
the
of
per
the
Of
cent
error
sample
interacts
only
and not vice versa.
In
second
following
sample
no
case
which
in
The
500
mg/IOOml
straight
line
fits
the
data
within
shown.
is directly
with
400
300
SAMPLE,
interaction.
concentration
tion of the preceding
sample
of the measured
sample,
and
en
200
PRECEDING
proportional
and is independent
that,
as a first
the
sample
was there
could
not
to
of the
approximation,
immediately
an observable
be attributed
the
concentra.
concentration
any
following
effect
to the
givit,
on the
follow-
252
THIERS
ing
one
sample
which
one can
oils-flow
and
plus,
preceded
of
assume
system
of its
glucose
3.1)
and
presumably
X
as
100
potassium
hut as
sample
immediately,
follows
space
sample
In
in
laboratory,
day
for
“noise”
Per
cent
I lists
taken
the
are
the
that
the per
widely
during
cent
any
given
day.
______
for
in some cases
this instrument
analyzed
of which
--
Table
Urea
one
interaction
one run
are
1.
that
selected
For
TYPICAL
Nvgar
given
reniams
(lay
a part
that
example,
if
source
Fig.
3 are
of
the
of
run
level
of
for the calibrafor each
day’scent
interactionsThese
the
samples
interdependwhich
Examination
examples
of
shows
method
does
not
constant
during
to day
AUTOANALYzER
-
this-
of tile
they
emphasize
similarity
of
for any
usually
from
not as
whether
week.
chemistry.
1 are typical
seen
of
checking
in clinical
of Table
and
[(3.6
day
location
like
randomly
case
the
be
be
preceded
the second
3.1still
show
the elevation
after
quite
large,
and
places
on the
together
the data
Variations
tests”
immediately
that
can
can
this
on
and before
running
standards
interactions
are then
calculated
typical
examples
of these
per
records
usually
records
and
measured,
sample,
directed
at
definitive
results.
method
the
Thus-
constituents,
period.
had
would
by
the continuof its urea
“in
a short
tray
result
sample
glucose.
through
portion
conditions
or after
sampling
4, the
Chemistry
The amount
in Fig.
4 and
interaction,
the
“interaction
each
instrumental
tion curve.
work.
Table
actions
are
ence which
cent
Under
Experiments
not shown
this
from
of its
that
for
sample.
shown
of each sample
was
a part
of the succeeding
the
Fig.
iii
due to interaction.
interaction
have
each
passing
definite
of each
“per
9%.
=
test,
9% of the
of that
sample
an empty
mEq./L.
of
obtained
as part
of the following
from
tests
such
as that
expressed
5.4]
--
Clinical
course,
any
enhancement
it. Similar
data
were
that
each
specimen
in
leaves
behind
a certain
this is measured
calculated
easily
conveniently
& OGLESBY
as
shown
by
vary
any
Table
U
(%)*
INTERACTION
Na
K
Cl
GO,
3.5
1.5
3.5
5.0
4.0
12
2.5
1.5
3.0
7.0
4.0
10
4.0
0.5
3.5
7.0
4.0
12
4.0
1.0
3.5
8.5
5.0
12
4.0
2.5
3.5
7.0
4.0
7.5
2.5
1.0
5.5
7.0
2.5
7.5
4.0
0.5
3.0
5.0
2.5
*Data
ured
(taken
constituent
from
of
routine
any
given
daily
sample
records
which
for
appears
1 week)
in
show
the
next
the
percentage
sample.
10
of
the
meas-
vol.
10, No.
3.
1964
INHERENT
Routine
correction
is not difficult.
Three
253
ERRORS
for errors
possibilities
due
to interaction
exist.
(1) Cups
between
of water
samples
can be
placed
after
each cup of standard
or sample.
This
makes
correction
for interaction
unnecessary
but it cuts
by 0110 half
the rate
at which
analyses
can he performed.
(2) If tile interaction
for potassium
oii a
day is 9c/ , theii
9% ot the result
for each
sample
can he sul)tracted
from
that the chart
sample
range-for
the result
on the following
peaks
are of the correct
which
correction
contains
potassium,
an aniount
Say 4.5
is necessary
for
sample.
(3) One can assume
height
only when
they
follow
roughly
nlEq./L.
samples
which
in the
tInier
middle
these
follow
samples
a
of the normal
conditions
no
a concen-
of
tration
between
follow
reasonably
close
to 4.5 mEq./L.
For
9% interaction,
sami)les
3.!) and 5.1 mEq./L.
C1I
be assumed
to affect
samples
which
theni
by less than
0.05 mEq./L.
For samples
of concentrations
higher
than
and
the
following
5.1
mEq./L.,
sample
sample.
difference
between
added
to the following
For
sional
of the
this
low
urea
and
samples
the
samples
must
lected
portion
2 is preferred
to
the
Table
2,
most
the
chloride,
reduces
to
daily
work
OF
divergent
AUTOANALYZF.R:
J)ala
(mEq./L.)
Corrected
(inEq./L.)
*Data
June
are
1962.
averages
of
of
9%
4.5
of
is
only
occa-
correction
particularly,
samples
which
in time.
saving
the
the
mEq./L.
render
of urea
and carbon
a relatively
correction
2 shows
data
load
and
dioxide,
small
of the
EFFECT
differences
results
from
fol-
the
Tn
Procedure
in
a
illustrates
data.
is altered
number
tile
marked
randomly
imse-
typical
Tt is noteworthy
by the correction
differthat
rep-
data.
OF CORRECTION
Na
Raw
to
many
in a considerable
interaction
Table
widely
l’RECISIoN
repeat
mEq./L.
because
high
enough
In tile case
ences
between
corrected
and uncorrected
the small
proportion
of the data
that
resents
and
4.5
result
he corrected.
of the
precision.
of
between
from
the
3.9 mEq./L.,
concentration
liecessary.
it unnecessary
resulting
potassium,
because
it
of
difference
subtracted
less
than
Procedure
case
of sodium,
3 is preferred
Application
provement
the
is
colIcelitratiolls
samples
makes
samples,
which
of
sample
sample.
glucose,
show
following
technic
uremic
9%
concentration
For
samples
K
FOR
1NTERACTION
Cl
1.73
0.10
1.62
1.38
0.96
0.08
0.81
1.05
between
183
sets
of
duplicates,
run
in
random
order,
254
THIERS
& OGLESBY
Clinical
Chemistry
Drift
In a continuous-flow
instrument
such
as the
AutoAnalyzer
which
depends
for precision
upon
the constancy
of fractional
completion
of
reaction,
drift
is a much
more
serious
inherent
problem
than
in analytical
for
methods
in general.
a particular
that
tile
tion
changes
result
of
determination,
relationship
ard
this
of
a progressive
problem
in
effect,
peaks
on
occurs,
that
a given
the
slowly
the
fashion
is
precision,
are
the
drift
strip
with
set
has
included
suffered.
of
To
at frequent
instrument.
each
day expressed
ing determined.
The
The
apparent
in terms
Drift
may
position
analyzed
in
not
precision
case,
and
source
of error.
Serum
samples
included
in the
from
series
to the
the
of the
periods
K
besame
or less random
fashis usually
negligibly
method.
may
be
ANY
of
constituent
in the
In other
methods,
the largest
single
standards”
in Fig.
IN
opera-
Table
3 shows
conditions.
The
1-hour
period
on
These
“drift-control
of correction
illustrated
DRIFT
1-HOUR
are
PERIOD
routinely
intervals
provide
5. This
(IEq./L.)*
#{163}1
COt
6
0.3
7
2.5
10
6
0.2
0.2
4
5
1.0
0.5
9
0.5
10
2.0
7
0.3
6
0.5
10
0.4
16
2.5
4
0.2
2
2.0
taken
from
routine
daily
records
for
1 week:
sugar
one
stand-
the
regular
ATJTOANALYZER
Na
known
a pool
of known
concentration
of analytical
samples
at
the procedure.
for the method
MAxIMUM
of the
drift
at
concentratwo sets
of
as distiii-.
concentration
drift.
routine
in any
long
so
concentra-
characteristic
drift,
change
of concentrations
continue
for
curve
throughout
small
as compared
however,
this
is
eData
and
evaluate
intervals
direction
in a more
glucose
methods,
drift
3.
its
chart
samples
direction,
or it may
change
ion.
Tn 1)0th the urea
and
Table
moves
time.
these
standards
is a measure
of instrumental
the amount
of drift
observed
under
normal
results
are given
as the maximum
drift
seen
throughout
the basis
calibration
the same
relationship
to each
other
in their
analyzed
at a different
time,
but that
the
be significantly
different.
Thus
accuracy,
from
samples
tion
between
in
time
will bear
tions
as when
answers
will
guished
When
and
urea
show
negligible
drift.
Vol.
10, No.
technic
accuracy
3,
1964
INHERENT
has proved
mandatory
is to be obtained
with
those
plate
for urea
is occupied
bration
curves
ginning
of
and
by
left-hand
drawii
40
of
portion
Fig.
shift
for
in our experience
AutoAnalyzers
glucose.
Every
a cup containing
are
a plate
of
5.
calibration-cue
drift
correction.
from
samples
Fig.
5.
is taken
centration
as
at that
is shifted
value
for
upward
the drift
the group
from
the
are
data
ence
in
the
elevation
time,
and
analytical
of
tile
since
the
this
technic
corrections
standards
tile strip
each
curve
as
as to pass
in question.
the
‘size
shown
operating
at
corrctiori
on
Accuracy
aspects
tinguished
that
Table
cotidrawn
4 shows
are
be seen
differnecessary
for day-to-day
on instruments
the
effect
of
this
at.
(7).
are
dis-
precision.
and Precision
Accuracy
Two
at
originally
through
tile correct
of the sami)les
in
course
of 1 hour.
This
technic
also corrects
in accuracy
which
have
been
observed
precision.
standa given
for
corrected
curve.
As can
can make
a considerable
of
bethe
All
during
the
differences
maximum
the
in
arbitrarily
drift-control
standard
curve
sampler
Cali-
at
chart
are
calibration
calibration
so
on the
standard.
samples
nearest
to
the drift-control
or downward
control
standard
accuracy
tenth
position
a drift-control
The
then
read
from
of Table
3, this
if even
reasonable
in all methods
except
the calibration
as shown
on
divided
into
groups
of samples
ard.
The height
of the peak
for
group
255
ERRORS
and
precision
have
of precision,
on the
basis
been
repeatability
that
the
latter
defined
and
includes
by
Hughes
et
reproducibility,
variability
due
to dif-
256
THIERS
Table
4.
PRECISION
OF
& OGLESBY
AUTOANALYZER:
Clinical
EFFECT
OF
CORRECTIONS
Chemistry
DRIFT
FOR
AND
INTERACTION*
Data
Na
Raw
(mEq./L.)
Corrected
(mEq./L.)
Data
run
in
are
standard
random
Table
5.
deviations
order,
TYPICAL
K
Cl
±2.9
±0.08
±2.2
±1.0
±0.9
±0.06
±0.9
±0.7
calculated
March
from
72
samples
from
(0,
six
different
serum
pools,
1962.
OPERATING
DATA
FOR
AUTOANALYZERS
DURING
Jurx
o
1962
Ai-ouwr
%
Determlnation*
Urea
Aeeuraeyf
nitrogen
Repeatabilityl
Interaction
-
1.0
3
-
2.7
1
+<0.1
Glucose
Reprodueibihtyl
+0.8
Drifts
-II
-II
Sodium
+0.9
1.1
2.3
3.5
7
Potassium
+0.05
0.02
0.15
7
0.3
+0.8
0.9
2.0
4
5
10
2
Chloride
Carbon
dioxide
*Except
and
for
I Average
Data
per
and
glucose
cent
interaction,
the
millie(luivalents
per
units
liter
are
for
milligrams
other
the
per
100
ml.
for
urea
nitrogen
determinations.
bias.
not
oh,tainable
froln
Expressed
as
IlNegligible
as compared
standard
#{182}Typical drift
per
ferent
operators,
similar
factors,
repeatability
described
performed
5 shows
present
routine
records.
deviations.
to precision
of
niethod.
hour.
different
(lays
whereas
can
terminations
bility
come
l)e
on one
only from
the
obtained
of operation,
former
by
pool
of serum
quality-control
does
one
as
the
as
the
average
repeatability
a standard
bias,
and
deviation,
discussed
are
employed.
interaction
characteristic
hourly
drift.
It
of
for
different
not.
samples,
Thus
technician
an
also
includes
these
methods
multiple
of
The
of
de-
reproduciprocedures
parameters
in this
methods
of methods
as
laboratory.
Table
discussed
here,
1-month
period.
It also
a typical
reproducibility
when
all
and
estimate
making
whereas
estimates
data
over time.
above
allow
estimates
of these
under
normal
routine
conditions
tile
accuracy
of the six automatic
expressed
shows
pressed
4
-
of
of
the
estimates
and
these
methods,
excorrection
methods
of
of the
their
average
average
Conclusion
Although
Analyzers
they
are
described
not
here
widely
recognized,
are
important
the
aspects
properties
of Autoof continuous-flow
Vol.
10.
No.
3,
analytical
fect under
mizing
INHERENT
systems.
routine
or
ments
1964
By recognizing
conditions,
technics
correcting
the
in piecision
rections
at
and
the
cost
errors
of very
them
and determining
have
been
developed
they
accuracy
little
257
ERRORS
produce.
result
from
extra
effort.
their
effor 1111111-
1)ramatic
improve-
utilization
of these
cor-
References
I.
4.
Technicon
AutoA
naly-er
Methodoloqy,
No.
75,
ments
Corp.,
Chauncey,
N. Y.
Technicon
A nbA
nalyzer
Methodology,
No.
16,
gen.
Tech,iicon
Instruments
Corp.,
Chauncey,
Tech nicon
AsitoA
nalyzer,
Booklet
511-5-60G.
N. Y., 1958.
Teehmnicon
Instruments
Corporation.
A utoinatie
5.
Marsh,
2.
3.
W.
Academic
6.
‘I’echnicon
H.
in
Advances
Press,
New
Instruments
in
Clinical
York
(1959),
Corporation
Electrolyte
Siniulboneou
hughes,
11. K.,
et al.
Anal.
Chem.
p. 338.
Announcer.
24,
1349
Instruments
Electrolytes
Chentistry,
(1952).
‘l’eehinicon
Glucose/Blood
N. Y.
Technicon
(1961).
7.
l’roee(lure.
I ustru-
Urea
Nitro-
(hum IIIuccy,
Corp.,
Bulletin.
vol.
2, edited
Tech.
Bull.
by
Req.
Sobotka
Medical
and
Tech.
Stewart.
31,
2