INDUSTRIAL a n d ENGINEERING CHEMISTRY
A N A L Y T I C A L E D I T I O N
VOLUME 18, NUMBER 10 ISSUED OCTOBER 21, 1946 CONSECUTIVE NUMBER 20
Assistant to Editor:
N .A .
PARKINSONManuscript Editing:
G . G LAD YS G O R D O N R. P. C H A P M A N . J. R. CHURCHILL B. L. CLARKEEDITOR: W A LTE R J. M U R P H Y Associate Editor:
LAW RENCET.
HALLETTAssistant Editors
Manuscript Reviewing:
STELLA AN D ERSO N Advisory BoardT. R. C U N N IN G H A M G. E. F. LUNDELL M . G. MELLON
Contributing Editor: R. H .
MULLERMake-up: CHARLOTTE C . SAYRE
R. H. MULLER B. L. OSER H. H. WILLARD
Use of Control Charts in A nalytical L a b o ra to ry ...
Grant Wernimont 587 Determination of Basic Nitrogen in Hydrocarbon Feed
Stocks...Leon Donn and Harry Levin 593 Use of High-Frequency Oscillators in Titrations and
A n a ly s e s ...F. W . Jensen and A . L. Parrack 595 Chromatography in Separation and Determination of Basic
A m ino A cid s . . . .M . S. Bergdoll and D. M . Doty 600 A id s to Computation in Spectrophotometric Analysis of
Binary M ixtures . . . .M ark Fred and F. W . Porsche 603 Conditions A ffecting Sequence of Organic Compounds in
Tswett A dsorption C o lu m n s ... H . H . Strain 605 Determination of Chlorine in 2,2'-D ihydroxy-5,5'-dichloro-
diphenylm ethane...
D. M . Jenkins, K. L. Waters, and G. D. Beal 609 Viscosities of Pure H yd ro ca rb o n s...
J. M . Geist and M . R. Cannon 611 Chromatographic Resolution of Q uinone O xim es. . . .
D. K. Gullstrom, H . P. Burchfield, and J. N . Judy 613 Determination of Free Fatty A cids in Dried Egg Powders .
C. M . Johnson and Leo Kline 617 lodometric M ethod for Assay of Penicillin Preparations . .
J. F. A licino 619 Spectrophotometric Changes during O xidation of Vitam in A
O i l s ... ...G. R. Halpern 621
The American Chemical Society assumes no responsibility for the statements and opinnSns advanced by contributors to its publications. Views expressed in the editorials are those of the editors and do not necessarily represent the official position of the American Chemical Society.Indirect Calibration of Filter Photometer by Means of S pectrophotom eter... C. L. Comar 626 Determination of Conjugated Diolefins with Chloromaleic
A nhydride . . .S. T. Putnam, M . L. Moss, and R. T. H all 628 Indicator Properties of Derivatives of 4,-Nitrophenylazo-1-
n a p h th o l...K. H . Ferber 631 Theory and Operation of Cartesian Diver Type of M ano-
s t a t ...Roger Gilm ont 633 Determination of Some A rohiatic Am ines and Substituted
Ureas in Smokeless P ow de r...
T. D. Waugh, Garman Harbottle, and R. M . Noyes 636 Analysis of Organoselenium C o m p o u n d s ...
J. D. M cCullough, T. W . Campbell, and N . J. Krilanovich 638 M ICROCHEM ISTRY
O xid e Films Formed on A llo y s at M oderate Temperatures E. As Gulbransen, R. T. Phelps, and J. W . Hickman 640 Preparation of Powdered Materials for Electron M icros
copy. . . . M . C. Schuster and E. F. Fullam 653 NO TE O N A N A L Y T IC A L PROCEDURES
Analysis of Boron Trifluoride in Organic Liquids (Ethers) . S. L. Walters and R. R. M ille r 658 C O R R ESPO N D EN C E ...658 Instrumentation in Analysis . . R. H . M tille r (A d vt. Sect.) 23
W e acknowledge with thanks the action of J. T. Baker Chemical Co. in releasing the front cover of this issue for.editorial purposes.Copyright 1946 by American Chemical Society.
37,000 copies of this issue printed.
Published by the American Chemical Society at Easton, Pa. Editorial Head
quarters: 1155 16th Street, N. W ., Washington 6, D. C./ telephone, Republic 5301/
cable, Jiechem (Washington). Chicago Editorial Branch: Room 819, 25 East Jackson Blvd., Chicago 4, III./ telephone, Wabash 7376. Houston Editorial Branch: 413 West Building, Houston 2, Texas; telephone, Capital 6516. N ew York Editorial Branch: 60 East 42nd Street, New York 17, N. Y ./ telephone, Murray H ill 2-4662.
San Francisco Editorial Branch: 24 California Street, San Francisco, Calif..
Business Office: American Chemical Society, 1155 16th Street, N. W ., Washington 6, D. C. Advertising Office: 332 West 42nd Street, New York 18, N. Y./ telephone, Bryant 9-4430.
Entered as second-class matter at the Post Office at Easton, Pa., under the Act of March 3, 1879, as 24 times a year— Industrial Edition monthly on the 1st, Analytical Edition monthly on the 15th. Acceptance for mailing at special rate of postage pro
vided for in Section 1103, Act of October 3,1917, authorized July 13,1918.
Remittances and orders for subscriptions and for single copies, notices of changes of address and new professional connections, and claims for missing numbers should
be sent to the American Chemical Society, 1155 16th Street, N. W ., Washington 6, D. C. Changes of address for the Industrial Edition must be received on or before the 18th of the preceding month and for the Analytical Edition not later than the 30th of the preceding month. Claims for missing numbers w ill not be allowed (1) if received more than 60 days from date of issue (owing to delivery hazards, no claims can be honored from subscribers in Continental Europe, Asia, or the Pacific Islands other than Hawaii), (2) if loss was due to failure of notice of change of address to be received before the dates specified in the preceding sentence, or (3) if the reason for claim is
"missing from files” . Annual subscriptions— Industrial Edition and Analytical Edition sold only as a , . . , . unit, members $3.00, nonmembers $4.00. Postage to countries not in the Pan- American Union $2.25; Canadian postage $0.75 Single copies—current issues, Industrial Edition $0.75, Analytical Edition $0.50/ back numbers, Industrial Edition
$0.80, Analytical Edition prices on request; The American Chemical Society also publishes Chemical and Engineering News, special rates to members.
Chemical Abstracts, and Journal of the American Chemical Socety. Rates on request
GOOD NUTRITION FOR AMERICA
. . . starts in the laboratory!
J. T. Baker Chem ical Co.,
Executive Offices and Plant:
Phillipsburg, N .J .Branch Offices:
New York, Philadelphia and Chicago.P u r i t y d e f i n e d — n o t t o ' ' m a x i m u m l i m i t s " — bu t to t h e d e c im a l b y a c t u a l lot a n a l y s i s . T h a t ’s t h e s t o r y of t h e B a k e r ’s A n a l y z e d l a b e l .
Som eone has said, " W h a t we eat we a re !” But few realize th at good h ealth frequently is charted in the laboratory!
Soils vary in chemical- content and so do crops th a t are grow n on them ! T he feed intake of livestock frequently determ ines the n u trien t values o f m eat, eggs, m ilk, etc.
V itam in and m ineral fortification o f food has been m ade possible th ro u g h laboratory analy
sis. L aboratory analysis has indicated n o t only the need fo r b u t also the degree o f enrichm ent necessary. T ruly, A m erica’s good h ealth starts in the laboratory.
In the laboratories (p ro d u ct control room s) o f m any o f the leading food and food process
ing m anufacturers, you w ill find effective tools to measure good nutrition.
T hey are Baker's Analyzed C.P. Chemicals and Acids.
T h e reason these Reagents have been selected is unique, for B aker’s Analyzed C.P. Chemicals bear this distinction. Each b ottle has the actual lo t analysis on th e label. T h e sm all percentages o f im p u r itie s by lo t a n a ly s is , as fo u n d by B aker’s analysts, are p rin ted on the label.
Chem ists w ho w ork in term s o f th ird and fo u rth decim al exactness know th e value of the actual analysis rath er th an a statem ent of m axim um lim its o f im purities.
So on your next order, be sure to specify B aker’s A nalyzed C.P. Chemicals and Acids.
Y o u r favorite laboratory supply house can supply you.
A N A L Y T I C A L ' E D I T I O N 5
Shown above on a background of the aberration curves w ith which American O ptical Com pany scientists predict lens perform ance, is a special optical system designed for use w ith an elaborate m achine tool.
M any such system s can be economically produced using stock optics from Spencer Scientific In stru m en ts. O thers m ust be specially designed. In either case, our long experience, large staff of optical specialists, and ex
tensive m anufacturing facilities are th e best assurance of satisfaction for th e m anufacturer who has problem s of an optical n atu re. W rite D ept. K48 for details.
American W Optical
C O M PA N Y
Scientific Instrument Division
Buffalo 15, New York
Predicted on Paper
6 I N D U S T R I A L A N D E N G I N E E R I N G C H E M I S T R Y Vol. 18, No. 10
CHICAGO APPARATUS COMPANY CH IC A G O 22,LA|LL IN O IS D esigned to meet the n eed of a ru g g ed instrum ent of
high optical quality to be used in the visual exam ination of opaque objects, polished metal specim ens, an d similar materials.
Supplem ented by a suitable cam era, it is useful in m aking perm anent photographic records.
In this model, features of design that make up a g en eral purpose instrum ent have b een incorporated. Am ple space has been provided for convenient m anipu
lation of all accessories.
The base is heavy and well balanced, con
tributing to stability.
Three types of body tube are avail
able and interchangeable, m onoc
ular, vertical and inclined b inocu
lar. In e ach case the tube length is 215 mm. and no chan g e of ob
jectives is req u ired for chang e of body tubes an d the m agnifications are essentially the same in e a ch
No. 76301 with m echanical stage sup
plied as a n . accessory at extra cost. W e are listing h ere two p o p u la r1 m onocular models as follows:
No. 76300 Model CM, M etallurgical Microscope, with vertical illum inator an d necessary objective handles, one each 8.0X, 20.0X and 37.0X achrom atic objectives, and one each 7.5X and 12.5X H uygenian eyepiece, complete m c a r r y m g c a s e ^ 3g gQ
No. 76301 Model CM, M etallurgical M icroscope, with vertical illum inator a n d objective handles, same as No. 76300 but with the addition of an illumi
nating unit, 6.5 volt, and a suitable resistance for operation on 115 volts A.C. or UX, . ,
com plete iA carrying case. Each ? 362-00
(Illustration shows this m icroscope with a m echanical stage w hich is not inclu ded m prices shown above.)
o rder for one of these instrum ents for delivery in two to three months.
No. E-223.
Both coarse an d fine focusing adjustments are provided.
The fine adjustm ent h ead is g ra d u ated in steps of 2.5 m icrons for convenience in estimating height of surface characteristics.
case.
P y r e x , V y c o r " a n d " C o r n i n g ” a r e r e g i s t e r e d tr a d e -m a r k s a n d i n d ic a te m a n u fa c tu r e b y
C O R N I N G G L A S S W O R K S • C O R N I N G , N E V/ Y O R K
P V R i X
F O R A L L - A R O U N D U S E . . . Y E A R R O U N D E C O N O M Y
brand LABORATORY GLASS WARB Research in Glass means --- P O R Y O U R C O N V E N I E N C E . . .
Scattered throughout the U nited States a n d C an ad a in convenient locations are stocks of Pyrex, Vycor and C orning b ran d s of laboratory glassw are. These are the inventories your laboratory supply d e a le r carries . . . carries for only one re a s o n —YOU.
In a sense e ach of these inventories is personal. For all a re different. Each is built-up an d m aintained for the selected group of custom ers served by a p a rtic ular laboratory supply house. Stocks c a rrie d are b ased on the d e a le r's know ledge of his custom ers' needs and a re always as com plete an d as well b a la n c e d as factory shipm ents perm it. If your d e a le r h asn 't a p articular item, you m ay be su re the factory h asn 't it either. That is why C orning Glass W orks urg es you to contact your d ealer w hatever your requirem ents may be. His is a professional service to professional men. H e is the one who c a n serve you best. The closer you work with him the more time an d m oney you save.
C orning Glass W orks is proud of its d ealers and wants all users of all its la b
oratory glassw are lines to lea rn to know their n earest dealers b etter . . . a n d to
use their services to the fullest extent.
i n d u s t r i a l a n d e n g i n e e r i n g c h e m i s t r y Vol. 18, No. 10 o o o o o o o o o o o o o o o o o o O O P o o o o o o o o o o o o o o o o o o o o o o o o OO o o o o o o o o o o o o o o
0 500 ml. S11.G5
0 1 liter 14.80
0 2 liters 16.70
0 3 liters 21.45
0 5 liters 26.25
o o o 0 0 0 0 0 0 0 V 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
9 *ftpsuw ed J ie a tiW f fja c k e t
GYCO H eating Jackets were developed to fill the need for a faster and more efficient laboratory heating device.
They are constructed of glass cloth and insulated with layers of glass wool.
GYCO heating elements are made of Safeway electrical resistance heating wire, sewn directly to the inner surface of the jacket. This intimate contact of the element with the vessel to be heated permits fast, economical operation and ease of constant control of temperature. The heating element can be operated below the visible glow point causing very little oxidation of the heating wire and no overheating of the insulation. Therefore, the life of the GYCO Heating Jacket is greatly increased.
GYCO Heating Jackets are made to fit all standard flask
sizes and special sizes can be made to order. They are also available in tubular units for use on columns of varying size.
Since hemispherical jackets are most widely used, GYCO Jackets are standardized in th a t form. As an accessory, jacket tops are available which convert any GYCO H eating Jacket into a spherical unit. These jacket tops are available with elements for extra heat requirements or w ithout elements for insulating purposes only.
All GYCO Jackets have a built-in iron constantan thermocouple ending in an “ AN” type plug for hooking up to the pyrometer of the GYCO Pyro-tran. (This plug m ay be removed for use with any other pyrom eter of similar type.)
CIEÏTIF 1 C CLASS
Listed below are a few sizes and prices. For a complete description of all models, accessories and prices, write for our Bulletin C-100.
GYCO Jacket Tops make it un
necessary to dismantle apparatus when removing jacket. They convert a hemispherical jacket into a spheri
cal and m ay be ordered separately.
G Y C O J a c k e ts are d is tr ib u te d e x c lu s iv e ly b y
APPARATUS COMPANY, Inc.
B lo o m fie ld , N ew J er se y
o o o o o o o o o o o o o o o o o o o o
WATER CHEMICALS
CONTROL-LAB TESTS
LEEDS & NORTHRUP
BOTAN Y M ETALS
L E E D S 4 N O R TH R U P COM PANY,
4920
STEN TON AVE., PH I LA., PA.Measu ring in s t r u m e n t s • t e l e m e t e r s • a u t o m a t icc o n t r o l s • h e a t-t r e a t i n gf u r n a c e s
M . Ad E-9G (25)
pH INDICATOR FOR "EVERYBODY"
H e re ’s th e p H In d ic a to r for th e m an w ho is not a p H e x p e rt.
I t ’s in te n d e d to he c a rrie d a ro u n d a n d used w h erev er d esired
— in p la n ts as well as labs. I t ’s as s tu r d y a n d d e p e n d a b le as a te m p e ra tu re in d ic a to r. I t will s ta y on th e jo b . T o use it, you ju s t m a k e 3 sim ple a d ju s tm e n ts , th e n p u t th e sam p le in to th e
beaker a n d see w here th e needle pointy.
“ S tic k y ” w e a th e r o r su rro u n d in g s w on’t affect th is In d ic a to r, unless re la tiv e h u m id ity is o v e r 95 a n d a m b ie n t te m p e ra tu re is over S5F. I t ’s fine for la b o rato ries, k itch en s, re frig e rate d chests, packing room s a n d pow er p la n ts.
S olution to be checked can be a t a n y te m p e ra tu re to 50C (120F). T h ic k so lu tio n s a n d “ s o ft” solids like b u tte r , cheese, dough a n d m a n y m e a ts, e a rth , e tc ., can be checked alm o st as easily as slurries or clear liquids.
N e a rb y electric al e q u ip m e n t w on’t affect th is in s tru m e n t. I t requires v e ry little a tte n tio n . I ts scale is s u b s ta n tia lly longer ancFéasier to re a d th a n in a n y co m p arab le In d ic a to r. C o n stru c - tio n ally , i t ’s a fine jo b . C o m p lete w ith e v e ry th in g necessary for pH m easu rem en ts.
F u r th e r d e ta ils are in C a ta lo g E-9(5(2), s e n t on re q u e st, b u t if you h av e a p H m e a su rin g problem we suggest you check w ith an L& N engineer.
I N D U S T R I A L A N D E N G I N E E R I N G C H E M I S T R Y Vol. 18, No. 10
I n t r o d u c i n g
The New INSTA-MOUNT
THE EMIL GREINER CO.
1 6 1 S I X T H A V E N U E ( ^ |||^ p N E W Y O R K 1 3 , N. Y.
O f l i c e , - 5 7 0 9 G R O V E S T R E E T , O A K L A N D , ” C A L . . 1 1 2 B R O A D W A Y , C A M B R I D G E , M A S S .
D e s i g n e d
A P P A R A T U S S U P P O R T F R A M E S to M e e t t h e S p e c i f i c D e m a n d s
L a b o r a t o r y I n s t a l l a t i o n s !
Illu str a te d h e re a re th e in te g ra l p a rts of th e In sia - M ount w h en d is a s s e m b le d . A n y b a r or c l a m p ,m a y be r e m o v e d w ith o u t d is tu rb in g th e to ta l a s s e m b ly
I N S T A - M O U N T S e t "A"
P ilot P la n t A s s e m b ly In sta -M o u n t c la m p s (d e m o u n t
a b le v is e s ) w a ll a n d floor fla n g e s to p ro v id e for th e a s s e m b ly of a 48" x 72" ch assis'. In sta -M o u n t S et " A ” m a y b e m o u n te d o n floor, wall or table. $ 6 4 .5 0
I N S T A - M O U N T S e t "B"
L ab A sse m b ly P ro v id e s a c o n v e n ie n t la b o r a to r y ta b le c h a s s is u p to 24" x 48".
In sta -M o u n t S et ’'B ‘ m a y b e m o u n te d o n w a ll or t a b le , c S h o w n h e re is th e v ise c la m p
f
>rinciple, u n iq u e w ith the n sta -M o u n t. The K e y w r e n c h , w h ic h fits in to flu sh m o u n te d s c r e w , p e r m i t s i m m e d i a t e tig h te n in g or c o m p le te d is a s s e m b ly o f unit.R e p r e s e n t i n g t h r e e im p o r t a n t a d v a n c e s in b o t h t h e d e s i g n a n d m a n u f a c t u r e o f a p p a r a t u s s u p p o r t fr a m e s , IN S T A -M O U N T is i n d i s p e n s a b l e for p ilo t p la n t a s s e m b l i e s a n d la b o r a t o r y s e t - u p s . C a p a b l e o f g i v i n g lo n g e r a n d m o r e s a t is f a c t o r y s e r v i c e , IN S T A -M O U N T o ffe r s
t h e s e m a j o r a d v a n t a g e s : 1. MAXIMUM RIGIDITY
O perating on a n e w v ise principle, in stea d of d e p en d in g upon uncertain screw strength, INSTA-MOUNT p rovid es 100 TIMES THE GRIP oi currently a v a ila b le laboratory clam p s.
2. MAXIMUM VERSATILITY
With INSTA-MOUNT, a set-u p is e a s ily a n d quickly m odified w ithout d ism an tlin g the entire framework, a s n e c e s s a r y w ith the old ty p e clam p. Each half of e a ch INSTA-MOUNT C lam p is in d e p e n d e n tly d is e n g a g e a b le — perm itting rem oval or relocation of a n y sin g le bar w ithout disturbing other bars a n d clam p s.
3; MAXIMUM SAFETY
INSTA-M OUNT C lam p s and Bars are in d ivid u ally p recision m a ch ined from 17ST Duralum inum . This affords m axim um strength w ith m inimum w eig h t. N o lo w m elting point a llo y s are u sed . At the sa m e tim e, flush m ounting of INSTA-MOUNT C lam ping S crew s m inim izes d a n g er oi g la s s b r e a k a g e b y protruding parts.
»^grpofatjgS»
October, 1946
MACBETH
< ' ' 1
3 1
■ T
m m ~ .
: s a
adSite&B&ai
T h e First
D irect R eading
Line O p erated
C o n tin u o us In d icatin g
■
( ______
p H M eter
The MACBETH pH METER protects production
—continuously
T h e M acb eth lin e -o p e ra ted p H M eter can be left o n co n tin u o u sly o r in definitely, as may be d esired . T h is is a fa c to r o f g re a t im p o rta n c e in m a in ta in in g a co n sta n t p H .c h e c k o n p r o d u ctio n . T h e la rg e , easily read scale, g ra d u ated in .1 p H u n its, c o n tin u ously fro m Oto 14 p H , gives in fo rm a tio n quick ly and clearly. U n tra in e d users can easily m ake o rd in a ry p H m easurem ents. T h e M acb eth is th e m o st sim p le p H M eter to o p e ra te .
MACBETH pH METER FEATURES ONLY ONE OPERATING CONTROL
Write j o r Bulletin.
a SBRhBBHHUB
MA C B E T H C O R P O R A T I O N
727 W E S T 1 7 t h S T . N E W Y O R K 11, N. Y.
/Manufacfurers of Scientific Apparatus Since 1915
'o-ttlidesi the M a ny AéAMÂAtUuj&i o j
WELCH STAINLESS STEEL TRIPLE-BEAM BALANCES
1) H igh S en sitiv ity — under all w o rk in g conditions, 2) C onsistent A ccuracy
3) Cob alite K n ife Edges
C overed — n o n -ru stin g
•
4) Covered A gate Bearings
•
5) O ne-Piece B eam C onstruction
s s a a r
6) 3 Etched Scales
C apacities 100 g .— 10 g .— 1 g-
visible at eye level
7) H ig h C orrosion
R esistance No.
4 030 P a te n t N o .1
,8 :S E N S IT IV IT Y : 0.01 g. or less a t to ta l c a p a c ity
C A P A C IT Y : 111 g. (w ith e x tr a w e ig h t 201 g.)
A b so lu te ly every m e ta l p a r t is of Stainless Steel, e x c e p t th e
b a se c a s tin g a n d pillar, w h ic h h a v e a c ry s ta l-fin is h coating.
October, 1946 A N A L Y T I C A L E D I T I O N
O a u im 1* '
for Particular Requirements throughout Industry
« K ° K K
< /e xa 7 > <(hormax)> ^ KIMBLE y
B L U E L IM E
C o n s u lt le a d in g L a b o r a t o r y S u p p ly H o u s e s th ro u g h o u t th e U n ite d S ta te s a n d C a n a d a , fo r K im b le L a b o ra to ry G la ssw a re to m eet YO UR needs.
W E L L K N O W N T H R O U G H O U T T H E W O R L D A S L E A D E R S I N D E V E L O P I N G A N D M A N U F A C T U R I N G
I N D U S T R I A L H E A T T R E A T I N G E Q U I P M E N T
S O L D E X C L U S I V E L Y T H R O U G H L A B O R A T O R Y E Q U I P M E N T D E A L E R S
I N D U S T R I A L A N D E N G I N E E R I N G C H E M I S T R Y Vol. 18, No. 10
The new Lindberg Volumetric T y p e C arb o n Determ inator incorpo
rates features which facilita te fa ster and more accu rate a n a ly se s of carbon in iron, steel, heat-resisting steel, stainless steel, etc.
The p recisely g ra d u a te d burette, mounted in front o f a fluores
cent light for rap id rea d in g s, offers an a ccu ra cy o f one point of carbon . . . or better. The leveling bottle is fre e fo r quick and e a sy leveling. Cups at lowermost and uppermost positions a re provided to hold bottle. A conveniently located micrometer screw in the low er cup allow s zero adjustment of the meniscus b efo re determ ination starts.
The contact typ e absorption cham b er permits complete absorption in two p asses . . . for many a llo ys one p ass is suf
ficient. G la s s tubing in the cham ber assures quick, thorough g a s dispersion. Burette is w a te r jacketed and g rad u a ted fo r 1-gram and '/4-gram sam ples, making the unit a d a p t a b le fo r full ran g e use.
The sturdy stand is attra ctively finished in glossy g re y b a k ed enam el. All g la ssw a re is P y rex to min
imize b re a k a g e . Panel mounting elim inates the n eed for clumsy clam ps.
Although designed for use with Lindberg Combustion Tube Furnaces, this new C a rb o n Determ inator is e a sily and quickly a d a p te d fo r operatio n with a n y stan d ard high tem perature combustion tube furnace.
S e e your Lab o ra to ry Equipment D e a le r to d a y for a d d i
tional d eta ils and prices.
L I N D B E R G E N G I N E E R I N G C O M P A N Y 2 45 0 WEST HUBBARD STREET • C H IC AG O 12, ILLINOIS
"T M u c A S o f t e n
T O B E A C C U R A T E
MALLINCKRODT
J 9 c j ?
M a llin ck ro d t St., St. Louis 7 , Mo
C H IC A G O • P H IL A D E L P H I A
CHEMICAL WORKS
¿ 0 ^ ê /m n u c a / f y f o e tâ
7 2 G o ld St., N ew Y o r k 8 , N. Y .
L O S A N G E L E S • M O N T R E A L
Vol. 18, No. 10
C R Y S T A L S T Y P I F Y P U R I T Y . . a n d M a l l i n c k r o d t A n a l y t i c a l R e a g e n t s a re tk e p u r e s t of cry stals. W h e n th e c h e m is t specifies o u r R e a g e n t s h e is a s s u r e d of h i g h e s t q u a l i t y a n d u n ifo r m , d e p e n d a b l e p u r ity .
S e n d for t h e M a l l i n c k r o d t A n a l y t i c a l R e a g e n t c a t a l o g to g e th e r w i t h a n y specific i n f o r m a t i o n d e s ir e d o n ^ M a llin c k ro d t c h e m ic a ls to fit y o u r s p e c ia liz e d o p e r a tio n s .
A lw a y s S p e c ify M a llin ck ro d t R e a g e n ts In O r ig in a l P a ck a g es
P otassiu m D icliro m a lc ( M allinckrodt) M agn ified to 10 D ia m eters
I N T E R N A T I O N A L M O D E L P R - 1
T h e la te s t d ev elo p m en t in R efrig erated C en trifu g es, th e In te rn a tio n a l M odel P R -1 offers the la b o ra to ry a n a ly st p ra c tic a lly all of th e a d v a n tages o f the larger p e rm a n e n t type in sta lla tio n s plus p o rta b ility . C e n trifuge and com pressor are com bined .in one a ttra c tiv e c a b in e t m o u n te d on caste rs, an d b o th u n its are o p e ra te d from a single cord an d plug w hich can be connected to th e o rd in a ry lighting circu it.
C o n sta n t te m p e ra tu re s plus or m in u s 2° can be m a in ta in e d , and u s a ble accessories illu stra te d here include th e m u ltisp ee d a tta c h m e n t and high-speed h ead s for six 7 m l. tu b es or four 25 m l. tu b e s a t 18,000 R .P .M ., conical angle h eads for 15, 50 an d 100 m l. tu b es a t speeds up to 5,000 R .P .M ., the four-place pin ty p e head for 250 m l. b o ttle s a t 2,600 R .P .M ., as well as th e co n v en tio n al horizo n tal tu b e ca rry in g heads. T h e com pressor is of am ple c a p a c ity to cool the in te rio r o f the g u ard bow l to 32° F . or lower w ith an y of this e q u ip m e n t o p e ra tin g a t m ax im u m speed an d a room te m p e ra tu re of 80° F.
A lth o u g h n o t previously an n o u n ced , th e M odel P R -1 h as a lread y been supplied to A rm y , N a v y and civilian lab o rato ries, an d has been su c
cessfully used in research on the c h em istry of the influenza bacillus.
T h e fe atu res of the m achine will a t once suggest countless ap p lic atio n s, an d com plete d e tails will gladly be fu rn ish ed on re q u est.
W idth 28" Length 44" H eight 42" W eight 850 lbs.
I N T E R N A T I O N A L E Q U I P M E N T C O M P A N Y
B O S T O N 3 5 , M A S S A C H U S E T T S
Vol. 18, No. 10
A.H .T. CO. SPECIFICATION
K O F L E R M IC R O H O T S T A G E
F o r d e te r m in in g c o rre c te d m ic r o m e ltin g p o in ts o n t h e m ic ro s c o p e s ta g e w it h s a m p le s
a s
s m a ll as a s in g le c ry s ta l
K O F L E R M IC R O H O T S T A G E (M ic ro M e ltin g P o in t A p p a ra tu s);
A .H .T . Co. S p ecificatio n , electrically h e a te d a n d w ith stag e c a lib ra te d th e rm o m e te rs.
For determining corrected micro melting points by means of a microscope with samples as sm all as a single crystal, permitting con
tinuous observation of changes in the sample before, during and after melting. Useful also for general micro-preparative work, sublima
tions, measurements of refractive indices at elevated temperatures, fusions, heating under controlled conditions, etc., and physico
chemical studies. See Ludwig Kofler, M ikro- chemie, Vol. X V (1934), p. 2 iS ; and Kofler, Kofler and Mavrhofer, “ M ikroskopiscne M eih o d en in d erM ik ro ch cm ie'' ( V ien n a, 1936).
For temperatures up to 350 °C, with an ac
curacy of ± 0 .5 °C in the range to 200 °C and of ± 1.0°C in higher range. Can be used with transmitted, reflected or polarized light on any compound microscope providing magnifi
cations from 50 to 100X w ith objective having working distance of 6 mm or more, and pref
erably with a metal stage.*
The apparatus consists essentially of an in
sulated, chromium plated m etal stage, 90 mm diameter X 20 mm high, heated by an em
bedded Ni chrome unit, and with a central light well fitted with a condensing lens system.
A threaded post takes either a fork for the micro slide or various sublimation blocks. A vertical sliding contact rheostat with rotary
6886-A . F ig . 1 C o m plete A ssem b ly
d S t S p Ä i l s ^ t e Ä p e c i a l l y designed for use with this H ot Stage, permits exact reproduction of
chart for the rheostat.
Fig. 2 G lass Baffle D in p o sitio n o n H o t S tag e
F ig. 3
T y p e B F isc h e r S u b lim atio n B lock in p o sitio n on H o t S tag e
F ig. 4
K o fler-D ^ rn b ach v a c u u m C h am b er in p o sitio n on H o t S tag e
Fig. 5
K o fler-D ern b ach V a cu u m C h a m b e r in p o sitio n o n F isc h er B lock
6886-A. M icro H o t S ta g e , K o fle r, as a b o v e d e sc rib e d , com pletei 'is s b o w J ^ ^ ^ b a f f l ^ D ^ h r e c ^ u b U m a t i o n d k h ^ K ^ K o f l ^ - D m i b a c h th e rm o m e te rs M , cooling b lo ck H , F isc h e r s u b lim a tio n b locks J a a , .g re a e e n ts T a n d v e rtic a l r h e o s ta t; in case, w ith
6887-A . D itto , H o t S ta g e A, only, w ith tw o c a lib ra te d th e rm o m e te rs M . cooling block H , glass baffle D a n d v e çh c a l rh e o s ta t. b u t ^ t h o u t ^ e or o th e r a ccesso ries... ... ... ...
* A v ailable on sp ecial o rd e r.
More detailed information sent upon request.
ARTHUR H. THOMAS COMPANY
R E T A IL — W H O L E S A L E — E X P O R TLABO RATO RY APPARATUS AND REAGENTS
WEST WASHINGTON SQUARE, PHILADELPH IA 5, PA., U.S.A.
C a b le A d d r e s s " B a la n c e ,” P h ila d e lp h ia
INDUSTRIAL and EN G IN EER IN G CHEMISTRY
P U B L I S H E D BY
T H E A M E R I C A N C H E M I C A L S O C I E T Y W A L T E R J. M U R P H Y , E D I T O R
Use of Control Charts in the A nalytical Laboratory
G R A N T W E R N IM O N T , Industrial Laboratory, Eastman Kodak Company, Rochester 4, N . Y.
Control charts are useful in the testing laboratory for comparing the over-all variability of test data with the average variability of small groups of the data, and are simpler to Understand than complicated statistical methods of analysis of variance. Operational meaning of control charts depends on the manner in which arbitrary small groups are chosen.
upon which control charts are based; and only a brief reference w ill be made to the simple arithmetical calculations involved because this part has been adequately presented in m any publi
cations ( /, 2, 6, 12-16). An attem pt w ill be made to show the operational meaning which can be put into control charts when applied to problems in the analytical laboratory.
S
T ATISTICAL methods have been used for a long tim e to present the average of test method results. However, as Fisher has pointed out (6), the variation among test results.usu- ally has not been an object of study, but has been recognized rather as a troublesome circumstance which detracted from the value of the average.The estim ates of test method precision which analysts do pre
sent are often not correct because they are made on the basis of a small amount of data covering only a short period of time. In many cases, a great deal of effort has been taken to eliminate all assignable causes of variation while the precision of the test method is being studied, even though it would not be desirable or even possible to do so when the test method is used for routine control purposes. W. S. Gosset, who published under the pseudonym of “Student” , stated all this aptly when he observed (17) that an analyst who wishes to impress his clients will ar
range to do repetition analysis as nearly as possible a t the same time, but if he wishes to diminish his real error he null separate them by as wide an interval as possible.
The application of statistical methods to industrial manu
facturing problems has been pioneered in the U nited States by Shewhart (14). During the war, the War Production Board ' sponsored some thirty-three
M A K IN G A CO N TRO L CHART
A control chart may be described as a graphic presentation of test data in such a manner that the variability of all the results is compared with the average variability within (arbitrary) small groups of the test results. The chart is said to show evidence of
“ control” when there is no more variation throughout the entire set of results than corresponds, statistically, to the average vari
ation within the (arbitrary) small groups.
In order to illustrate the preparation of a control chart, use can be made of a comprehensive experiment made by Shewhart (14), in drawing marked chips from a bowl. In his experiment, 998 chips were marked as shown in Figure 1, and 1000 groups of four chips were drawn w ith replacement from the well-mixed bowl. The averages and ranges (difference between highest and lowest) of the first fifty drawings are plotted in Figure 2. The grand average was —0.08 and the average range of four was 2.03.
It can be seen that this graph of random drawings is similar to a graph for routine test results, in th at both the averages and ranges fluctuate and the question im m ediately arises as to whether any limits for these variations can be set up.
Table I can be used to set up these lim its as follows: The upper short courses on the industrial
application o f S h e w h a r t ’s methods in which alm ost two thousand men were trained.
With the coming of these
“quality control” methods into chemical-process manufactur
i n g , it will be increasingly im
portant for analysts to be able to determine how much of any observed process variation is to be ascribed to the test pro
cedure itself. The control chart method of Shewhart is prob
ably the most practical method of studying the precision and accuracy of routine test meth
ods, and several authors have already mentioned its applica
tion to the study of chemical test procedures (4, 7, 8,1 0 ,2 0 ).
N o attem pt will be made here to develop the mathematics
Table I. Factors for Computing 3-Sigma Control Limits
N o.
of O b- ser- v a - tions,
n 2 3 4 5 6 7 8 9 10
C h a r t fo r S ta n d a rd
D e v iatio n s C h a r t for R an g e s C h a r t fo r In d iv id u a ls , F a c -
to rs fo r C o n tro l L im its A v. s ta n d a rd
d e v ia tio n , 5 .3 2 4 .1 5 3 .7 6 3 .5 7 3 .4 5 3 .3 8 3 .3 2 . 3 .2 8 3 .2 5
A v. ra n g e ,
2.66Ij
1 .7 7 1 .4 6 1 .2 9 1 .1 8 1 . 1 1 1 .0 5 1.01 0 .9 7
C h a r t fo r A verages, F a c to rs for C o n tro l L im its A v . s ta n d a rd
d e v ia tio n , A i 3 .7 6 2 .3 9 1.88 1 .6 0 1 .4 1 1 .2 8 1 .1 8 1 .0 9 1 .0 3
A v . ran g e,
At
1.88 1.02 0 .7 3 0 .5 8 0 .4 8 0 .4 2 0 .3 7 0 .3 4 0 .3 1
F a c to r for c e n tra l
line,
ci
0 .5 6 4 2 0 .7 2 3 6 0 .7 9 7 9 0 .8 4 0 7 0.8686 0 .8 8 8 2 0 .9 0 2 7 0 .9 1 3 9 0 .9 2 2 7
F a c to rs for c o n tro l
lim its
Bx 00 00 0
0.10 0 .1 7 0 .2 3 0 .2 7
3 .6 6 2 .6 9 2 .3 3 2 .1 3 2 .0 0 1 .9 0 1 .8 3 1 .7 7 1 .7 3
F a c to r fo r c e n tra l
line,
di
1 .1 2 8 1 .6 9 3 2 .0 5 9 2 .3 2 6 2 .5 3 4 2 .7 0 4 2 .8 4 7 2 .9 7 0 3 .0 7 8
F a c to rs for c o n tro l
lim its D% D<
00 00 0
0 .0 8 0 .1 4 0 .1 8 0 . 2 2
3 .2 7 2 .5 7 2 .2 8 2 . 1 1 2 . 0 0 1 .9 2 1.86 1 .8 2 1 .7 8 C o n tro l lim its fo r in d iv id u a l o b serv atio n s «
C o n tro l lim its fo r a v erag e s of g ro u p s of n C o n tro l lim its fo r s ta n d a r d d e v ia tio n s C o n tro l lim its fo r ra n g e s =» R ± 3<r^
— Cl3 "
" X * Vn«
0 *
■ I \ T - X 3 R
' dx
Ai<r
X * U R
3R
\ / ndi X A xR
\ / 2 n c
= D iR to D*R
a =* Bi j to B i /
X “ g ra n d a v erag e ; a = a v erag e s ta n d a rd d e v ia tio n ; R « a v erag e ran g e
W ith th e ex cep tio n of fa c to rs in co lu m n s fo r I \ a n d Ix, th is ta b le is ta k e n from ( / ) T a b le I, p ag e 50; th e fo rm u las a r e e x p lain ed on pages 52 a n d 53.
587
588 I N D U S T R I A L A N D E N G I N E E R I N G C H E M I S T R Y Vol. 18, No. 10 and lower limits for range variation are equal to the average
range, 2.03, multiplied by the factors 0 and 2.28 (which are found in columns D3 and D, corresponding to a group size of four under chart for ranges). This gives a lower lim it of 0 and an upper lim it of 4.6 for the variation of ranges of four drawings.
Ranges greater than 4.6 (and less than 6.0) are possible but they should be observed only one or tw o times per thousand drawings by chance alone.
Figure 1. Distribution of 998 Chips in Shewhart’s Sampling Experiment
Average, 0.0 Standard deviation, 1.007
In a similar manner, the lim its of variation for the averages of four about the grand average arc equal to the average range, 2.03, multiplied by the factor 0.73 (which is found in column A 2, corresponding to a group size of four under chart for averages).
This gives the limits —0.08 =*= 1.48 for the variation of averages of four drawings. A summary of these calculations is given in Table II.
Figure 2. Two Hundred Drawings from Shewhart's Bowl of Chips in Groups of Fours
These so-called “control lim its” are drawn in as shown in Fig
ure 3, and it can be seen that the ranges and averages are all within their control limits, as of course they should be, if Shew
hart’s experiment was carried out properly. The control chart tells us graphically that there was no more variation among the fifty averaged values than corresponds, statistically, to the aver
age range of four individual chips drawn at one tim e. If an in
finite number of drawings were tabulated and summed up, the results would make a distribution curve like th at shown a t the right o f Figure 3; the control lim its correspond, approximately, to the points on either side of the grand average where this dis
tribution curve approaches the zero base line.
The table of factors can be used to estim ate control lim its for
Table II. Summary of Calculation of Control Limits for Shewhart’s Drawing Experiment
aw ing N o. X i Ar2 Arj X R
1 1 .7 0 .2 1 .4 0 .5 0 .9 5 0 1 .5
2 0 .3 0 .3 0 .4 0 .4 0 .3 5 0 0 .1
3 0 .4 0 .4 0 .1 0 .4 0 .3 2 5 0 .3
4 • - 1 . 8 - 0 . 9 - 0 . 6 1 .7 - 0 . 4 0 0 3 .5
5 0 .5 - 0 . 7 0 .8 1 .0 0 .4 0 0 1 .7
46 Ôlô 2 2 " à - Ô . 2 61650 2 .4
47 0 .5 - 0 l 7 - 1 . 9 - 2 . 5 - 1 . 1 5 0 3 .0
48 0 .3 - 1 . 5 1 .0 - 0 . 1 - 0 . 7 5 2 .5
49 - 1 . 2 0 - 0 . 8 - 0 . 5 - 0 . 6 2 5 1 .2
50 - 0 . 7 0 - 1 . 0 - 1 . 1 - 0 . 7 0 0 1 .1
Sum - 4 . 0 0 10 1 .4
\ A v. - 0 . 0 8 2 .0 3
C o n tro l lim its fo r R — (0 X 2.03) to (2.28 X 2.03) = 0 to 4.6 C o n tro l lim its fo r X = - 0 . 0 8 =*= (0.73 X 2.03) => - 1 . 5 6 to + 1 .4 0
F o r c o m p lete d a ta s e e ,(14) T a b le A, p. 442.
ranges, standard deviations, and averages of small groups of test results as well as for the individual results themselves. When these lim its are drawn into the graph, the resulting control chart shows a t a glance whether there is more variation among the groups (or among all the individual results) than corresponds, statistically, to the average variation within the groups. If ex
cessive variation is found to be present, the control chart will often help to find its cause.
CHECKING PERFORMANCE OF ROUTINE TEST METHODS M any laboratories maintain a “ controlled sample” which is used at regular intervals to check the performance of a routine test method. The control chart in Figure 4 shows such test results for a simple viscometer. The controlled sample is run every day just before the production samples are tested. The five weekly results for the controlled sample are grouped in order to estimate control limits. The lower graph for weekly ranges indicates no variations greater than are to be expected bÿ chance alone, and the upper graph shows th at there is no more variation among the individual results from week to week than corresponds, statistically, to the average variation of any one week. These limits of test variation are satisfactory to the production depart
ment.
This kind of chart is invaluable when new test operators arc being trained or when it becomes necessary to change the con
trolled sample. T he chart also stops effectively most of the
•arguments betw een-the testing laboratory and the production department over the validity of production test results which are not as they should be.
Sometimes it is possible to keep a satisfactory check on a rou-
Figure 3. Control Chart for Averages of Four Drawings from
Shewhart's Bowl of Chips
tine test method w ithout the use of a controlled sample. Figure 5 shows a control chart for the moisture content of production dis
tillation lots of an organic solvent. The lower graph shows that the differences between two consecutive lots of the solvent do not fluctuate more than corresponds to chance alone. The upper graph for individual lots indicates that there is no more variation from lot to lot than corresponds, statistically, to the average dif
ference between two consecutive lots. This control chart com
bines the variations of the production department and the testing laboratory, but the chart still serves a useful purpose in the labo
ratory as long as all results arc within their own control limits and the control lim its are within the specification lim its (in this case 0 to 0.6% moisture).
Figure 5. Control Chart for Routine Moisture Determinations on Production Lots of an Organic
Solvent
manner which is both objective and quantitative. Figure 6 shows the results of a comparison between the so-called method of single swings and the more commonly used multiple-swung method for weighing on the analytical balance.
Four individual comparisons of two 1-gram and two 100-gram weights were made on ten different days using the two methods.
With one exception, the daily variation (standard deviation) does not fluctuate significantly greater than it should by chance alone and, with four exceptions, the individual weighings are all within their control limits.
N o assignable causes were found which would account for the occasional lack of control for individual weighings, but they m ay be the result, of air drafts or faulty manipulation in releasing the beam rests. A control chart for daily averages would indi
cate some evidence of more variability in the weighings from one day to another than corresponds to the average daily vari
ability. However, there is no evidence that either of the two weighing procedures is superior to the other for weighing loads up to the full capacity of the balance. One im portant thing to note in favor of the method of single swings is the fact that, it requires about one quarter as much time as the multiple-swing method.
M ention has been made of control charts for individual test results and for averages of test results. I t is always better, from a statistical point of view, to chart averages rather than the indi
viduals them selves. However, there are m any occasions in the analytical laboratory where single observations rather than averages are used or reported. Thus, only rarely would weighings be made in replicate and the average w eight used or reported.
In such cases the individuals can be charted, although they must be grouped arbitrarily in order to estim ate control limits. It is often desirable to chart averages as well as individuals because the chart for averages gives valuable information about the general pattern of a set of test results, even though it is only' the individual results which are reported or used.
C O NTRO LLING ERRORS OF C ALIBRATIO N OF TESTING EQUIPMENT
The calibration of testing equipment is often plagued with the errors of the calibration method itself, and control charts will offer help in such cases. Figure 7 shows the results of calibrating nine burets by weighing the water delivered a t a known tem perature to the 39-, 40-, and 41-ml. marks. The lowest graph shows the range of three consecutive individual tests made for each mark, and it can be seen that there are no variations from range to range greater than correspond to chance alone, although the upper lim it of 0.03 ml. seems high. These range variations were not the result of insufficient drainage tim e, and they seem to represent real differences in the amount of water held up on the walls of the buret.
When the routine test results show lack of control, the analyst checks his • reagents and then verifies his results on replicate samples from the suspected lot. The use of such a control chart som etim es makes it possible to reduce the amount of routine testing when a production process is not giving trouble.
As soon as lack of process control is indicated, the amount of routine testing can be increased immediately.
C O M PA RIN G MERITS OF A LTER N A TIVE TEST PROCEDURES
It is often desirable to • compare the merits of alternative test procedures, and a con
trol chart will usually help to do this in a
- ONE GRAM W EIG H TS •
i - 0 2 - S IN G L E SW IN G S
t "
uj -J ♦ 0 I CD 2 3 ? 5 2£ 5 5 * -02 -
m I 0 0 6 -
M U L T IPL E SW INGS -
- O N E HUNDRE0 GRAM W E IG H T S — - S IN G L E SW IN G S - - M U L T IP L E SW IN G S -
W 4 i V / W ri I t -
0 0 4
00 2
o L.i l. i l i i ,i i i i i i i ! i : : s i —i—L—;
OAYS 2 4 6 0 10 2 4 6 8 10 2 4 6 0 10 2 4 6 8 10
Figure 6. Control Chart for Comparing W eighing M ethods
590 I N D U S T R I A L A N D E N G I N E E R I N G C H E M I S T R Y Vol. 18, No. 10
Figure 7. Control Chart for Buret Calibrations
Each buret checked at 39-, 40-, and 41-ml. marksFigure 8. Control Chart for H eat of Combustion of
CP.Benzene
Data of Dean, Williams, and Fisher (3)
Figure 9. Control Chart for a Calorimeter Constant
The control limits for averaged corrections have been drawn about zero as the mean, in the upper graph, because all corrections would be zero if the burets were perfectly marked. There is more variation among the corrections than corresponds, statistically, to the average range of three consecutive individual calibrations.
Therefore, the corrections are applied if th ey are greater than
=*=0.01 ml. but no correction is justified if it is ± 0 .0 1 ml. or less.
In other words, corrections which fall within the control limits based on the variability of the method of calibration are not significant and no gain in accuracy results from applying them.
IM PO RTANCE O F PROPER GROUPING O F TEST RESULTS The manner in which test results are grouped is very important, and the next two control charts illustrate this very well. Figure 8 shows the results of tests on the reproducibility of a calorimeter over a period of about 30 days (S). The graph for differences between consecutive days is satisfactory, and the daily indi
viduals show about the same variability as that corresponding to the average difference between consecutive days. The results are outside control lim its on two days and there are too many results approaching the control limits. This indicates that there m ay be more variation from day to day than corresponds to the variation in any one day and the data, as collected, will never show it up.
Figure 9 shows a chart for the calibration of a fuel calorimeter, in which replicate calibrations were made on the same day.
D aily standard deviations and averages are again all within their control limits, which, in this case, means that there is po more variation from day to day than corresponds, statistically, to the average variation on any one day. The two charts shown in Figures 8 and 9 illustrate how the same type of experimental work can be made to have different operational meanings merely by carefully choosing the method of grouping the tests.
Another illustration of the importance of picking the proper arbitrary small group is shown in Figure 10. A controlled sample of sodium hydroxide in potassium hydroxide was sub
m itted for routine analysis twice a m onth for 18 months? The authors (21) treated the entire set of results as a single group and concluded that the test method was satisfactory. However, when the data are plotted, it seems evident that the monthly range and the individual determinations them selves show less variability w ith time. I f the data are divided in the middle, two periods are obtained during which there is no more variation from month to month than corresponds, statistically, to the average m onthly differences; but the lim its are reduced approximately one half during the second period. I t is possible th at some change in the test procedure was made early in 1942 which will account for these improved results. In any event, had a chart such as this been in use prior to August, 1942, the authors might have found an assignable cause for the two low results of that month.
Figure 10. Control Chart for Determination of Sodium H ydroxide in a Potassium H ydroxide Control Sample
Data of Williams and Haines (2 /)
October, 1946 A N A L Y T I C A L E D I T I O N
METHOD POTASSIUM 01-PHTHALATE C. B HYDROCHLORIC ACIO G LA C IA L ACETIC ACIO
Figure 11. Control Chart for Sodium H ydroxide Standardizations
The application of control chart m ethods to data already col
lected'often shows that the same amount of experimental work would have yielded a great deal more useful information if it had been planned in a slightly different way.
materials, or in the analysts, and the control chart should help materially in locating the cause. N o con
clusions about the relative merits of the three primary standards can be safely drawn until the cause of this lack of control is removed.
The control chart in Figure 12 summarizes data for nine methods of determining lead (9). Known solutions containing exactly 70.0 mg. of lead were tested by each method with four replicates at the same time in each test. The m ethods were all different, five gravi
metric and four volumetric. The lower graph shows that the range fluctuations are no greater than correspond to chance alone, although the authors expressed the opinion, based on other considerations, th at method 5 showed excessive variation. The control limits for averages were drawn about 70.0 mg. rather than the grand average of all the results because the controlled . samples all contained 70.0 mg. of lead. Three of the methods gave excessively low results, based on the average range of the nine tests.
'The control chart in Figure 13 summarizes the results of a collaborative study of gravimetric and volumetric methods for determining sulfur in coal (18). The sulfur was determined as
C O M PARISO N O F C O LLA B O R A TIVE STUDIES O F TEST METHODS An -obvious application of control charts is to the comparison of the results of the collaborative study of test methods, and several examples will be presented to show this. Figure 11 shows the results of such a study of the standardization of 0.1 N solutions of sodium hydroxide (11). Unknown samples of the solution were standardized in twelve different laboratories by means of three primary standards. The lower graph shows the range of three individual standardizations made a t the same tim e, and the upper graph shows the average of the three. This control chart indicates th at there is some uncontrolled factor which af
fects the ranges of some of the analysts more than corresponds to chance alone, and there is also more variation among the analysts than corresponds, statistically, to the average range of the indi
vidual standardizations made b y each analyst. Three labora
tories are consistently out of control and one laboratory gave bad results when using one of the standards. The cause of these abnormal variations m ight be in the standardizing procedure, in th e controlled sample of sodium hydroxide, in the standardizing
Figure 12. Control Chart for Comparing Various Test M ethods for Lead
Data of Lykken (5)
M ETH O D BOM B E S C H K A T E T R A - SODIUM
WASHING H YD R O X Y- R H O D lZ -
QUINQNE ONATE
Figure 13. Control Chart for M ethods of Determining Sulfur in Coal
Data of Tomkins (18)
barium sulfate in both gravimetric methods, but in one case calorimeter bomb washings were used while in the other separate samples of coal were carried through the Eschka method. Bomb washings were used for both volumetric methods but different indicators served to detect the end point of the final sulfate titra
tion with barium chloride.
The lower graph for range of each collaborator’s individual determinations shows th at the gravimetric methods are appre
ciably less variable than the volumetric methods and that in both cases the ranges do not show more variation than corresponds to chance alone. One might conclude from this th at the volumetric methods are less satisfactory than the gravimetric methods.
This is not the case, however, because the upper graph for aver
ages shows that the collaborators cannot agree as well using the gravimetric methods as they do using the volumetric methods.
Unless the cause of this lack of agreement between collabora
tors using the gravimetric methods can be found and removed it would be better to use the volumetric methods even though they are somewhat less precise. N othing can be said about the ac
curacy of the methods because the sulfur content of the controlled samples of coal was not known.
