40 



Indiana University Studies 



journey, truck, executive, motor, gloves, argument, berth, mittens* 

 boots, summons, pants, dynamo, national, shirt, dress, opinion, 

 gown, witness, kimona, veil, drill, press, brooch, hammer, 

 belt, decision, pulley, barrette, tourist, judgment, tie, shaft, plea, 

 collar, code, skirt, piston, valve, legal, apron, anvil, sweater, 

 representative, pilgrimage, law, sledge, jersey, planer, prosecute, 

 waist, thermostat, vest, brief, defendant, petticoat, muffler, 

 thermometer, mayor, gauge, guilty, scarf, derrick, trip, prosecutor, 

 muff, boa, torts, channeller, closure, plume, recommit, cap, 

 bonnet, advisement, ulster, alibi, counsel, cuff, majority, frock, 

 luggage, guy, insignia, crane, uniform, minority, evidence, 

 livery, allege, pardon, raiment, carburetor, apparel, penitentiary, 

 badge, boiler, prison, indictment, convict, baggage, supreme, 

 costume, battery, tour, federal, helmet, vat, circuit, habit, 

 state, excursion, cowl, mill, theft, appeal, hood, injunction, 

 shawl, jig, garb, arraign, hardAvare, tax. 



Tests of this type covering about 1,000 words and 27 classes 

 were given to grades 3 to 8 in 14 schools. The tabulated returns 

 cover the reactions of between 1,500 and 2,500 children to each 

 word, or a total of 1,600,000 classifications. Under the caption 'Tn- 

 diana-Thorndike Vocabulary Scale", 774 of these words are here 

 arranged in groups, each group representing a distinct measure 

 of difficulty. The zero point is arbitrarily assumed to be 5 steps 

 below the easiest step on this scale. The units of difficulty are 

 in terms of P. E. distances on the base of a normal probability 

 surface. The method of scaling is that used by Thorndike in 

 constructing the original scale. The entire group of words yields 

 a scale of 16 steps, each step being about .3 P. E. 



Altho in listing the words for this scale, the average error 

 for all the children of all the grades has been taken as a basis, 

 it is an open question whether words can be selected from so wide 

 a range successfully. It might be better to consider a more limited 

 range of grades. We might make a third grade scale based on the 

 returns from the third grade, a fourth grade scale on returns 

 from the fourth grade, etc. Certainly for the first and sec- 

 ond grade tests, the reactions of first and second grade children 

 are of much greater importance than the returns from any higher 

 grades. However, the advantages of having a single scale to ap- 

 ply to a wide range of classes is great, and it seems worth while 

 to try to make one on that basis. 



It is then assumed that when words show the same average per 

 cent of error they are of eqilal value. But in selecting from the 

 entire scale the words for Scale R and Scale R2, two further cau- 

 tions have been kept in mind. First, words should show a de- 



