August, 1957, Vol. XXXIX, No. 8 583

Breaking the Subject Index Barrier—

A Citation Index for Chemical Patents*

BY EUGENE GARFIELD*

I

It is not my intention, as the title of my talk mightimply, to attack the excellence of the various indexingservices. We all know hew indispensable they are.When I speak of the subject index barrier, I have in mindthe intellectual limits of a perfectly conceived and exe-cuted subject index, wherein the subject relationships ofthe knowledge recorded is dependent upon the interpre-tation of authors and/or indexers and must of necessityfail to anticipate all the polyvalent viewpoints of users.Every thought, idea or discovery can take on new mean-ing depending upon the user’s frame of reference. When$as is often the case, the frame of reference of author orindexer is the same as that of the user then conventionalsubject indexes are quite useful.

The indexer, by judicious arrangements of biblio-graphical terms, can bring out aspects of a work that theauthor would overlook. He may also place relatedtirticles in closer juxtaposition by editing the index afteral~ entries have been collimated. Keeping the user inmind, the conscientious indexer will translate the termi-nology and phraseology of the author into a standardizedand more usabIe form, However, the indexer is, of neces-sity, primarily guided by the subject content whichauthors provide.

The indexer also faced with a practical economic bar-rier cannot index with the almost infinite depth to befound in the (3tation Index. The Citation Index breaksthis “barrier” by presenting subject matter in Biblio-graphical arrays which are neither alphabetical norclassified but associative.

● Presented at the Minneapolis meeting of the American Chemical Sossi-ety September 16, 19S5.

T Director, The fnstitute for Scientific fnfonnation, 1122 Spring ~essStreet, Philadelphia 23, Pema.

Reprinted from the Journal of the patant Office Society, Volume 39, Number 8,August 1957, PII. 583-595.

472

584 Journal of the Patsmt Ofice society

II

At this point it would be well to describe a CitationIndex, Even those familiar with Shepard’s Citations 1may not be able to immediately visualize a similar systemfor use in patent or literature searching. A CitationIndex for patents might indicate under the serial numberfor each patent those subsequent patents which had madereference to it. An example might look as follows:

Patent No.1,970,5782,437,2612,450,0792,451,1492,~,3492,518,2302,541,285

DateA&. 21, :34

Sep. “ 2 ‘%Oct. 12, ’48Mar. 7, ’50Aug. 8, ’50Feb. 13, ’51

ChemicaIAbstractsReference2$, 6332(7)

44, 5m9i44, 11,040i45, 4S92d

InventorC. SchodlerL. L. LernerK. R. BrownE. BoehmM. DeGrooteL. FreedmanA. L, Rawlins

Pa&t.

204582dM97.2&.;4

m~o

167-81

Figure 1

The citing patents could be classified chronologically,by inventor or by principal subject matter. This samplelisting is by no means complete. It is pertinent to citeat the end of a patent, references to the published litera-ture, as well as issued patents. Indeed, it is often just asvaluable to begin a search with a literature referencerather than a previously g“i’anted patent. However, I

have limited the scope of this paper to issued patents andpatent citations.

HI

In conducting research the scientist must be interestedin information that is either in anticipation of his ownproposed methods or ideas or information which con-firms them. Anyone who has done some research willremember that small batch of reprints you kept foldedin your lab coat—you neater fellows used looseleaf note-books, or verticle file folders. These reprints were thefew “gems” that you had found when you were still inthe preliminary stages of your experimental work and

1 Adair, W. C. Citation Indexes for scientific literature. Amer. Docu-mentation 6, 31-2 ( 1955).

473

August, 1957’, Vol. XXXIX, NO. 8 585

thinking. Quite often these reprints were so frequentlyused they did not physically survive the research pro-gram. They constituted your bible for a while, and un-less you were doing something completely unrelated toany thing previously reported in the literature-a rarephenomenon—it contained a good deal of informationrelating directly to your immediate interests. Duringthis time any work bearing on the subject of this reprintor having the slightest relevance was extremely impor-tant, as a single new idea could mean a whole new lineof experimentation. In a certain sense this basic reprintor set of reprints was your base of operations—andbibliographically speaking you might wish that the uni-verse of the index were built around it-—a sort of biblio-graphic solar system.

I have labored this point as a means of poititing outthe principal feature of the Citation Index. The accesspoints in conventional indexes are subject headings, to-gether with their sub-headings, and modifications. In thePatent Office the access point is the sub-class, the sub-class name being comparable to the subject heading inan index. In the Citation Index the access point is theindividual patent or article, or more accurately the ideaor ideas which it presents, implies, or which one caninfer. Through the Citation Index one learns of thebibliographical descendants of any individual patent orarticle, of those contributions which for a variety ofreasons have cited the work in question. By referenceto the patent Citation Index subsequently issued patents(and articles ) that had referred to the patent in questioncould be instantly located.

Needless to say such a bibliographical tool, if current,would permit the scientist or inventor with very special-ized interests to keep posted on confirmations, emenda-tions, and refutations of his own published patents orarticles as well as those of his colleagues. These biblio-graphical descendants may be entire articles or they maylie portions of articles which in their general context areLtnrelated to the principal subject matter of the citedpatent. This is particularly true when a specialized

474

586 Journal of the Patent Ofice society

technique or procedure is being used in two or more pat-ents that are seemingly unrelated, Iwhere the proceduresare thereby unknown to the chemist who is possibly in-terested in the application of the technique and, particu-larly some small modification in it, which may not attractthe attention of the indexer. The old technique may alsocome up in a new patent claiming the technique for a new+compound. A further ramification of this particularpoint is the help the Citation Index would offer the in-ventor or his firm in keeping abreast of issued patentswhich on further investigation are not valid. Examinersdo not always have enough time to completely evaluatethe very patents they have cited. The conscientious pat-ent man interested in protecting his client and havingan intimate knowledge of the patents already owned byhis firm can observe more quickly the prior disclosure orproper scope of prior patents.

IV

There are two types of references in patents. Thereare those that are similar to references found in theliterature that are provided by the inventor in his appli-cation and disclosure. These will not be the concern ofmy subsequent remarks. However, the references toprior patents made by the patent examiner in his re-sponses to the inventor’s claims is a more interestinggroup of citations. These may or may not result in thedisallowance of the inventor’s claims. In any event theycontain pertinent subject matter found in the examiner’ssearch. These references are to be found in all patentspublished since February 4, 1947 based on an order is-sued December 19, 1946 that they be printed at the end ofthe claims section of patent specifications. Prior to thattime references of record are only to be found in thepatent “wrapper.” These citations are used a greatdeal in searches and at times an old wrapper will beexamined simply to locate these references of record.

Unlike the indexer, the examiner has the legal point ofview as well as the subject approach and therefore hasthe ability and is required to draw conclusions from whatis apparent in a disclosure and also what is implied. In

475

August, 1957, vol. XXXIX, No. 8 587

his response he often “stretches” his imagination somew-hat in interpreting prior art as being anticipatory.Under the duress of time the examiner may do this withthe idea of leaying it up to the applicant to find the mist-ake in his reasoning. The examiner will often be look-ing for a combination of ideas and will have to draw onpatehts in several subclasses to. make his point. In sodoing he has “indexed” those patents in a way that anindexer could not. His frame of reference is quite differ-ent when searching the prior art than it is when classify-ing claims.

One of the greatest advantages of a Citation Index isthat it brings together patents that are seemingly unre-lated, i.e. unrelated in terms of their principal subjectmatter. The closer two patents are in their obvious sub-ject matter the less chance that either one would beoverlooked in a routine search. Hence, it wou;d be de-sirable to organize a Citation Index for all classes ofpatents so that related patents in different main classescould be found. In order to properly illustrate this ad-vantage one would have to compile a complete CitationIndex. The example shown in Fig. 1 was drawn from asample based, on a file of 5,000 accessible chemical pat-ents. It is not a completely random sample nor is theexample a complete Citation Index entry. However, eventhis incomplete sample will illustrate the point.

The principal patent in question is one granted toSchoeller in 1934 on” Wetting and Dispersing Agents foruse in the Textile Industries. ” This patent was origi-nally classified in class 260-98. It has since been re-classified to 260-458, because of the revision of class 260in 1938.

This class covers patents on carbocyclic or acycliccarbon compounds which are esters (and processes ofmaking same ) of the sulfoxy acid type but limited toacyclic sulfuric acid esters.

This particular patent was subsequently cross refer-enced at the patent office to a multitude of sub-classes in-cluding the following:

476

588 Journal of the Patent Oflce Socwty

8-34 , 93 , 127 , 138, 139167-91 ,252- 8.7, 8.8, 8.9,356,357260-400, 404, 410.6,468,488,563,584, 614

On the other hand the CA abstract for this patent wasindexed under “dispersing agents” and under “wettingagents. ” There was no organic compound indexing.

The patents which subsequently referred to Schoeller %patent are shown. The first patent was issued to Lernerin 1948 on “Condensation products of Cholesterol Esterswith Polyethylene Glycol & Process for ProducingSame. ” The principal class for this patent is 260-397.2,sterols (including Vitamin D). It was cross-referencedin 252-354. It should be noted that neither of theseclasses corresponds to any of the numerous sub-classesin which the cited patent was placed. It was not possibleto completely determine how this patent was indexed byChemical Abstracts. It was not indexed under the head-ing sterols. It was indexed under “Cholesterol Esters”and under ‘‘ Glycols, Polyethylene. ”

The next patent was issued to Brown the same year on“Waxy polyol ether-esters. ” Its principal class is260-234, Carbohydrate esters. It was cross-referenced in260-210 and 260-410.6, Sub-class 210 is glycosides.Neither the cited patent nor the Lerner patent fall inthis class. The sub-class 410.6 covers ‘ (syntheticallyproduced higher fatty esters with acyclic polyoxy alco-hols. ” This sub-class subtends the Schoeller patent, butnot the Lerner patent. This same patent was indexedunder the following in CA: Waxes, Ethers, Esters,polyol, Hydroxy compounds, Palmitic acid, Stearic Acid,Glycerol, Sorbitol, Mannitol, D-Glucose, Ethylene oxide,propylene oxide, lubricants, cosmetics, yarns, sizes,polishing materials, and coatings, plus some umleter-mined organic compound headings.

I don’t think it is necessary to labor this point anyfurther. A surmpary of this information will be foundin the appended table. It is not intended here to criticizethe classifications assigned to these patents by the PatentOffice, nor the indexing performed by CA. The sample

477

August, 1957, Vol. XXXIX, i’?o. 8 589

does demonstrate, I believe, that a Citation Index willbring together related patents, the principal or apparentsubject matter of which are unrelated.

In order to determine more precisely how these patentsdo relate an examination was made of the patent wrap-pers in question at the Patent Office in Washington. Therecords of the Lerner patent show that the examiner didnot find the Schoeller patent in his search since theclasses he searched did not subtend this patent. Hefound it in his own files or memory or elsewhere. Hestates, in his response to Lerner’s original application‘‘ Schoeller describes condensation of various acids in-cluding stearic acid and its ester with polyethyleneglycols. ‘ ‘

In the wrapper for Brown’s patent the examiner states‘‘ Schoeller discloses the reaction of a carboxylic acid anddiethylene glycol and ethylene oxide. There is no inven-tion in substituting a trihydrici alcohol, or a similar polyolin place of the dihydric alcohol. ” Examination of thesecomments demonstrates the relationship between thethree patents involved, the frame of reference in this in-stance being condensation of acids with polyethyleneglycols. Thus, the Lerner and Brown patents werebrought together in the Citation Index through theSchoeller patent. One might start with the Lerner patent(found in CA under polyethylene glycols), find theSchoeller patent as a reference at the end of the specifica-tion, and through the Citation Index pick up the Brownpatent. (One must always remember that patents arenot classified by the subject matter of the disclosure butrather according to the claims that have been granted.However, cross references will be made to certain search-able disclosures in the patent specification). Citationsare provided for a variety of reasons, but principally todisqualify certain claims. Citations are also made fre-quently in order to restrict the scope of the application.The examiner will refer to a patent whose scope he thinkscovers the subject matter in question, Other citationsmay be used during dkcussions of “legalities.

478

590 Jowvwl of the Patent Oj%ce Society

Citations in patents often include references to theliterature or CA abstracts. The utility of a CitationIndex including these citations would be increased buithat would be the subject of a more detailed presenta-tion.z

1 do not intend to make any suggestions at this tim’econcerning the possible compilation of a complete Cita-tion Index to patents. Some time ago a formal proposalwas submitted to the Patent Office, suggesting that afeasibility study be conducted.s If anyone is interestedthey can question the Patent Office on the matter. Pend-ing action by the Patent Office, which seems doubtful, Iam hopeful that Chemical Abstracts and/or Shepard’sCitations may be able to undertake” the compilation of aCitation Index to patents. As recently noted in C&ENCA has indicated that Citation Indexes will one day bestudied by their new research group.4 About ten yearsago Mr. Harry C. Hart approached Shepard’s Citations.Both he 6 and subsequently Mr. Arthur Seidell a commu-nicated with the Patent Office again to no avail. Both ofthese gentlemen presented the viewpoint of the, PatentAttorney.

The lack of action in the Patent Office is particularlyunfortunate since a system of this type for the exclusiveuse by the examiners could include references to aban-doned applications which are not classified at all andwhich contain a wealth of citation information insofar asthe citations resulted in abandonment. It is not unlikelythat many searches have been repeated by examiners inpart or in whole when an examination of the appropriateabandoned file would turn up much useful data. Exami-ners do often remember these abandoned files. However,

ZGarfield, E. Citation’ Indexes for science. Science 122, KELl I (1955).a Though the patent CMce considered the citation Index a basically good

idea, it was concluded that the time required to compile and use it wouldnot justify acceptance of the proposal. The author does not agree withthese conclusions and hopes that future research will resolve this difference

%“’%E~&s’%, 2752-4 (1955).e Chemical Abstracts Service-Good Buy or Good-by.

S Hart, fiarry C Re: “Cltatlon System for Patent Office. j. Patentoff. SOL31, 714 (1949).

e Seidell, Arthur H. Citation System for Patent Office. J. PatentCy. Sot. 31, 554 (1949).

479

August, 1957, vol. XXXIX, No. 8 591

it is difficult to see why they have to rely “on their ruexno-ries for abandoned files and not for published patents.Classification is used precisely because the humanmemory has its limits.

I anxiously seek your comments and criticisms andhope to present data in the future as to how this tech-nique can be used to advantage even in small patent files.

I will be glad to elaborate on machine methods avail-able for compiling a Citation Index during the discussionperiod if there is time.

In closing I should like to mention that Miss MargaretCourain, Supervisor of the Research Files Division atMerck prepared the thousands of citation cards requiredin this experiment—and that without her work this datawould not have been obtained.

I also want to thank Dr. Charles Bernier who helpedin providing information on CA indexing and has mademany valuable suggestions. I am also indebted to Mr.Harold Lindenmeyer who helped obtain some of the in-formation on Patent Office cross-references.

480

592 Journal of the Pated O#ice Society

Pat. Off

Title Numbe[ Date Inventor Class&Cross Ref.

Nos.

Assistants for the Tex- 1,907,578 Aug. 21, 1934 G. %yller 26(WJtile and Related In-dustries M. Wittwer 404

410.*

563%4

614

167-91

&HJ

127

138

I39

252- 8.7

8.8

8.9

3%

357

Condensation products 2,437,261 Mar. 9, 1948 L. Lerner ;60:O;:.2*of Cholesterol Esterswith PolyethyleneG[ycol and processfor producing same

481

A%gwt, 1957, ~Or?. XXXIX, No. 8 593

Subject Matter of Examiner’s CommentsSub-Class**

CA Indexing

*Main Class

Acyclic sulfuric acidesters

Sulfoxy containing fatsNitrogen containing fatsSynthetically produced

higher fatt esters withl“acyclic po yoxy

alcoholsCarboxylic acid estersCarboxylic acjd esters,

acycllc acetic ser]es,unsubstituted

AminesAmines, acyclic, oxo or

OxyEthers, acyclic

Cosmetics, skin, creamsand lotlons

Vat or sulfur dyesMisc. dye compositions,

with dispersing agents,orgamc, oxy or hy-droxy comp~s. or theirsulfur analogs

Chem. modification oftextiles and fibers, mer-cerizing, wetting ~ddi-tion or pretreatment

Cleaning or laundering,degummlng or desizing

Cleaning or laundering,scour!ng, decreasing orbowk[ng

Textile treating, org.sulfoxy containing

Textile treat]ng, org.amine containing

Textile treating, ethercontaining

Colloids, wetting agents,protein or carboxyliccompd. containing

Colloids, wetting agents,org. amine, amide orN-base containing

**See entire classifica-tion schedules forcomplete descrip-tion of sub-classes

Dispersing agentsWetting agentsCA 28, 6332(7)

Sterols (including vit. D) Schoeller describes Cholesterol estersWetting agents, org. sul- condensation of va- Glycols Polyethylene

foxy, . protein or car- rious acids includ- CA 4.2,461cboxyllc containing ing steoric acid and

its ester with Poly-ethylene Glvcols.

482

594 Journal of +he Patent Oflce Society

Pat. Off

Title Number Date Inventor Class &Cross Ref.

Nos.

Waxy Polyol Esters 2,450,079 Sept. 28, ]948 K. Brown 260-210234*410.6

Manufacture of materi- 2,451,149 Oct. 12, 1948 Boehm 167- ~als resistant to 0[ ac-tive against mlcro- 32organisms

Process for breaking 2,5(K),349 Mar. 7, 1~ DeGrootePetroleum emulsions

252-34@260-410.6

A ueous solutions of 2,518,230 Aug. 8, 1950~ipoid-Soluble vita-

Freedman 167: ~~*

reins

A ueous solutions of 2,541,285 Feb. 13, 195I Rawlins%itarnin A esters

167- 81*

483

August, 1957, vol. XXXZX, NO. 8 595

Subject Matter of Examiner’s CommentsSub-Class** CA Indexing

●Main Class

Carbohydrates, glycosides Schoeller discloses the Waxes; ethers; esters,Carbohydrates, esters reaction of a car-See above boxylic acid and di-

polyol; hydroxy ac-ids; Palm.itic acid;

ethylene glycol and Stearlc acid; Glyce-ethylene oxide. rol; %rbitol; Man-There is no inven- rutol; D-gluc~;tion in substitutinga trihydric alcoho? ~~~~~e oxi%~~or a similar polyol bricants,; cosmetics,in place of the dihy- ~arn, SJzes, polish-dric alcohol. mg materials, coat-

2%, 8B6d

Insecticides, etc., car- See example 10 in pat- Disinfectantsboxyclic ent to Schoeller et Ethylene Glycol

Insecticides, etc., car- al., and particularlyboxydiic, phenolic

CA 4J, 352d

Insecticides, etc., car-the disclosure ofEthylene Glycol

boxyclic, napthalene Monocresyl Ether.

Resolving colloids, liquid N~th~;~:pecific about Emulsions, Petroleum,primary dispersants, Stearic acid, S;$e;gorg. agents, protein or (cyclic);carboxyllc contamlng acid, esters; Pen-

See above taerythritol (estersof Stearic acid);Polypentaerythri-tols (esters ofStearic acid)—nogreen marking

CA 44, 50B91

Vitamins S., who shows appli- Vitamins A, D{ Ecants compds. used No green markmgfor dispersing and CA 44, 11,040isolubilizing pharma-ceutical agents. Nornvention is foundin applyin

!these

ether comp s, of S.for solubilizing fat-soluble vitamins.

See above Rejected as lacking Vitamin A, B2, C, B6,invention over S. BI, E, K; Esters,who shows appli- nicotinamide Pant~cants esters used as thenic acid (Nadispersing. “and solu- Salt); Stearic acidbillzing agents for (monoester withfats and oils in Polyethylene Gly-pharm. industry. To co]) ; gl ycols, Poly-“employ the S. ester ethylene (estersas a solubilizer for of): Palmitic Acid;Vitamin A esters Arachidjc Acid;would not involve Margarlc, Linolelcinvention. and Oleic Acid

CA 45, 4892d