1

Classes and Sets

BUILDING SENTENCES

Before introducing the basic notions of set theory, it will be useful to make
certain observations on the use oflanguage.
By a sentence we will mean a statement whict in a given context is
unambiguously either true or -false. Thus
London s the capital of England.

'

Money grows on trees.

Snow is black.

are examples of sentences. We will use letters P, Q, R, S, etc., to denote

sentences; used in this sensg P, for instance, is to be understood as asserting
that "P is true."
Sentences may be combined in various ways to form more complicated
sentences. Often, the truth or falsity of the compound sentence is completely
determined by the truth or falsity of its component parts. Thus, if P is a
sentence, one of the simplest senteqces we may form from P is the.neqation of
P, denoted by - P (to be read "not P"), which is understood to assert that
"P is false." Now if P is true, then, quite clearly, - P is false; and if P is false,
then - P is true. It is convenient to display the relationship between - P and
P in the following truth table,
1.1

where and/denote the "truth values", true andfalse.

Another simple operation on sentences is conjunction: if P and Q are

thWUignction of P and Q, denoted by P nQ (to be read "P and Q"),
is understood to assert that "P is trw and Q is true." It is intuitively clear that
PnQ is true if P and Qarcboth true, and false otherwise; thus, we have the
following truth table.
sentences,

22

Classes and Sets

1.2

TheJSjgfionf P and Q denoted by P v 0 (to be read "P or Q"), is the

senten which asserts that "P k true, or Q k true, or P and Q are both true."
It is clear that P vQ is false only if P and Q are both false.
1.3

An especially important operation on sentencet is impli"ution jf P and

Q are sentenceq then P - Q (to be read "P implies Q") asserts that "if P is
true, then Q is true." A word of caution: in ordinary usagg "if P is truq then
0 is true" is understood to mean that there is a causal relationship between
F and Q @s in "if John passes the course, then John can graduate"). In
in the
thematics, however, implication is alw4y urlllq$
P + Qis
In
otherwords,
lse.
P is true a
P+Oistruee
the truth table.
1.4

The properties of formal implication differ somewhat from the properties we

would expect "causal" implication to have., For example,

"I + | :2 :>

z is a transcendental number"

is true, even though there is no causal relationship between the two compohent
sentences. To take another example,

"2 + 2 :

5 +

4isaprimenumber"

is trug even though the two component sentences are false. This should not

disturb the reader undul for formal implication still has the fundamental
if P - Q is trug then,

property which we demand of implication-namely,

necessarily, if P is true then Q is true.

Building Sentences

23

Certain compound sentences are true regardless of the truth or falsity of

their component parts; a typical example is the sentence P > P. Regardless
of whether P is true or false, P + P is always true; in other words, no matter
what sentence P is, P + P is true. For future reference we record a few
sentences which have this property.

1.5 Theorem For all sentences P and Q the following statements are true.

i) P + PvQ.
ii) PnQ-P.

i)' Q-PvQ.
ii)' P nQ + Q.

Proqf
i) We wish to prove that if P and Q are any sentences, then P + PvQ is
true; in other words, we wish to prove that no matter what truth values
are assumed by P and p, P - P v Q is always true. To do this, we derive
a truth table for P + P v Q asfollows.

P,
t
t

f
f

"f

PvQ
t
t

.f

t
t

P+PvQ
t

t
t
t

The basic idea of the derived truth table is this: in line 1, P and Q both take
the value ; thus, by 1.3, PvQ takes the value t; now, P has the value I
and PvQ has the value /, so, by I.4,P + PvQ takes the value /. We do
the same for each line, and we find that in every line.(that is, for every
possible assignment of truth values to P and Q) F
- P v Q has the value
(true). This is what we had set out to prove.
i)' The derived truth table for Q + P v Q is analogous to the one for
P + P v Q; the conclusion is the same.
ii) ln order to prove that P nQ- P for all sentences P and Q, we derive a

truthtableforPQ-P.

PQ

PnQ+P

In every line (that iq for every possible assignment of truth values to p

and Q), P nQ+ P takes the value /; thus, P nQ+ p is true irespective
of the truth or falsity of its cornponent sentences P and e.

Classes and Sets

24

ii)'

The truth table for P Q- Q is analogous to the one for Pn

the conclusion is the same. I

Q+ P,and

1.6 Theorem For all sentences P, Q and R, the following is true:

l(P

Proof

Q)

^(Q-

R)l

(P

R).

The reader should derive the truth table for

l(P-Q)n(Q+R)l + (P+R)
and veriff that this sentence takes the truth value r in every line of the

1.7 Theorem For all sentences P, Q and R,

1f

table'l

R is true, then

i\ PvQ-PvRist{ue,and
ii) PnQ-PnRistrue.
Proo-f

i) we

assume that

Q+

Pla

ii)

R is true, and derive the truth table for P v Q

lRl

PvQ

>

P v R.

PvQ+ PvR

Since we assume that Q + R is true, we cannot have, simultaneously, Q

true and R false; thuq we may disregard the sixth line of the table. In all
of the remaining lines, Pv Q+ PvR takes the value '
The proof that P rQ + P R is analogous to the above.

We agree that P

Q is to be an abbreviation f9r (P

Q\

x(Q

P\'

1.8 Theorem For all sentences P, Q and R, the following are true:

i) Pv Q+ Qv P,
ii) Pv(0vR)+(PvQ)vR,
iii) Pn(0 v R) +
iv) PvP+P,

(P n Q)v(P nR),

i)'PnQ+Qt'P,
ii)'Pn(QnR)e(PnQ)nR,
iii)' Pv(g nR) +. (Pv Q)n(PvR),

iv)'PnP+P.

The proof of this theorem is left as an exercise for the reader.

Building Classes

In

25

this and the subsequent chapters,

+ will be used as an abbreviation for

will be used as an abbrwiation for and. onry
if
rtwe will sometimes
write "iff" instead of +), n will be used as an
abbreuiuiio' f", ;;;';;i";
will be used as an abbreviation for or. rf p,
Q, R, ... are any statements, an
expression of the form p
e => R => ... should be understood to mean that
P - Q, Q = R,and so on; analogously, p o
eoR e... should be under_
stood to mean that p o e, o R,and-so on.
e
As is customary, r is to be read there exists, y is
to be readfor au, and is
to be
implies,

read such that.

EXERCTSES 1.1

Prove Theorem L8.

Prove that the following sentexces are true for

all p and g ioevrorgan,s Laws)

a)

-(PvQ)+-p-e.

-(PnQ)+-pv-e.

b)

3. Prove that the following sentences are true, for

every sentence p.
a) --P-P.
b) P+ --P
4. Prove that the following sentences are true for p

+ Q)o(-e + -p).
c) (P + Q) o -(P -e).
a) (P

e) [(PvQ)n

-p]>

e.

all

b)

and e.

(P+ Q)+(pve).

d) [Pn(P

5. Prove the following sentences are true for

a) [(P + 0@
- R)] =, (p - R).

a1l

p,

Q)]

e.

and R.

!{l - 0)n(R + 0)l + [(pvR ) - e]

[(p> e)(p-R)]e[p-(0^Rf
6. Prove that, for all sentences p, and R, if +R
is true, then the following
e
e
a) pv e <> pvR.
b) p xe+ pR.
c) (P + 0)o (p * i).
7. Prove that for all sentences p,
e, R and S, if p + e andR > S. then
a) Pv,( + QvS,
b) PnR + Qn,S.
b)
c)

are true.

BUILDING CTASSES

We will now begin our development of axiomatic

set theory.
E"1y axiomatic system, as we have seen, must start
with a certain number
of"undefined notions. For examplg in geomeiry,
the words ..poi.rt,;-urro ..line,,

and the relation of "incidenc"l' *" g"enerally

taken to be undefined. while
in our own minds to attach a "meaning,,, in
the form of a mental

we are free

26

Classesand Sets

picture, to each of these notions, mathematically we must proceed ..as ,,

if we
did not know what they meant.
en "unrtefinert" notinir has no
rrolerfie3
.Nnrv
are explicit
to it: t
axioms all I
ele
which we ex

to have.

In our system of axiomatic set theory we choose two undefined notions:

the word class and the membershp relation e. All the objects
of our theory are
called classes. certain classes, to be called sers, will be defined in
section 7.
Every set is a class, but not conversely; a class which is not a set
is called a

proper class.
Let us comment briefly on the "meaning" we intend to attach to these
notions. In the intended interpretation of our axiomatic systerq the word
c/ass
is understood to refer to any collection of objects. However, as we noted
in
section 2 of chapter 0, certain "excessively large" collections can be
formed in
intuitive set theory (for exampre, the collction of all x such that x
I x), and if
we do not exercise special caution they lead to contradictions such
as Russell,s
paradox. The"term proper c/ass is understood to refer to these ..excessively
largie" collections; all other collections are sers.

If x and ,4 are classeg the expression x e ,4 is read ..x is an element of

A,,,
or "x belongs to .4," or "x is in A." It is convenient to write x A fot ,,x is
$
not an element of A." Let x be a class; if there exists a class ,4 such that
x e A,

then x is called an element.

From here on we will use the following notational'conventlon: rower-case
letters eb, c, x, !,...will be qsed only to designate elements. Thus,
a
let{er, such as A,may denote either an element or a class which is
"upi;;l
not an element,
but a lower-case letteq such as x, may denote only art element.

A andB be classes; we define

B to mean that every
element of ,4 is an element of B and vice versa. In symbols,

1.9 Definition Let

A:B

A:

itr

xeA+.xeB and xeB+xeA.

we have defined two, classes to be equar if and only if they have

the same
elements. Equal classes havc another property: irx and y
ar equal and x is
an element of A, we certainly expect y to L an element
of . This property
is stated as our first axiom:
.A1.

Ifx :

yand x e A,then y e A.

This axiom is sometimes called the axiom of extent.

1.I0 Definition Let Aand B be classes; wedefine A

element of ,4 is an element of

B.

In symbols,

B to mean that every

Building Classes

'

AeB

lf A e B, then we say that A

We define

itr

27

xeA+xeB.

is a subclass of B.

A e B to mean that A c B

A is a strict subclass of B.
If A is a subclass of B, and .4 is a

set, we

and

t'

B; in this casg we say that

will call A a subset of B.

A few simple properties of equality and inclusion are given in the

next

theorem.

1.11 Theorem For all classes A,

and C, the following hold:

1) A: A.
11)A:B+B:A..
nDA:BandB:C>A:C.
iv\ A c BandB c A> A: B.
v)AcBandBeC+AcC.

Proqf
i) The statement

xeA-xeAand xeA+xeAis obviously true; thus,

A : A.
ii) Suppose A: B; then x eA+ xeB and xB+xeA; hence by 1.8(i)'
x e B + x e A and x e A > x e B;thuq by Definition I.9, B : A.
iii) Suppose A : B and B : C; then we have the following:
xeA+,fi,
xB+,E1,
by Definition 1.9,

:Z',liZ';

From the first and third of these statements we conclude (by 1.6) that
x e A + x e C. From the second and fourth of these statements we
concludethatxe C+xe ,4. ThuqbyDefinition 1.9,A: C.
We leave the proofs of (iv) and (v) as an exercise for the reader.

We have seen that the intuitive way of making classes is to name a property
of objects and form the class of all the objects which have that property. Our
second axiom allows us to make classes in this manner.

!4.

Let P(x) designate a statement about x which can be expressed entirely

in terms of the symbols , v, A, -, *, 3, V, brackets, and variables x, !, z,
A, B, ... Then there exists a class C which consists of all the elements x which
satisfy P(x).

2A

Classes and Sets

Axiom 42 is called the axiom qf

class construction.

The reader should note that axiom A2 permits us to form the class of all
tbe elements x which satisff P(x),not the class of all the classes x which satisfy
P(x); as we discussed on page 13, this distinction is sufficient to eliminate the
logical paradoxes. The semantic paradoxes have been avoided by admitting
in axiom A2 only those statements P(x) which can be written entirely in terms
of the symbols e, v, A, r,
3, V, brackets and variables.
The class C whose existence is asserted by Axiom A,2 will be designated

*,

by the symbol

{x lrlx}.

l.l2 Remurk The use of a small x in the expression {x I f(x)} is not accidental,
but quite essntial. Indeed, we have ugt..d that lower-cas l"tters x, y, etc.,
will be used only to designate elements. Thus
c

{*lp(r)}

asserts that C is the class of all the elements

x which satisfy p(x).

We will now use the axiom of class construction to build

from given

,or. ,r"* classes

classes.

l.l3 Definition Let A andB be classes ; the union of A andB is defined to be

the class of all the elements which belong either to A, or to B, or to both ,4
and B. In symbols.

AvB: {rl" eAorxeB}.

Thus, x e

v B if and only if x e A or x e B.

l.l4 Definition Let and B be classes; rhe intersection ol Aand B is defined

to be the class of all the elements which belong to bth A and B. In symbols,

AB: {xl xeAandxeB\.

Thus, x e A B if and only if xe AandxeB.

1.15 Defnition By the unitersal class ll we mean the class of all elements.
The existence of the universal class is a consequence of the axiom of class
construction, for if we take P(x) to be the statement x : x,then M guarantees
the existence of a class which consists of all the elements which satisff x : x;
by 1.11(i), every element is in this class.

Building Classes

1.16 Definition By ihe empty c/ass we mean the class @ which has no elements
at all. The existence of the empty class is a consequence of the axiom of class
construction; indeed, M guarantees the existence of a class which consists of
all the elements which satisfy x * x; by Theorem 1.11(i), this class has no
elements.

1.17 Theorem For every class -4, the following hold:

i)=..

Ii)Acolt.

Proof

i) In order to prove that @ c A,we

must show that x

e@+xeA.

suffices to prove the contrapositive of this statement, that is, x

Well, suppose x

x(A+x4.

(A+

It

x O.

A; then certainly x 4 , for @ has no elements; thus

ii) If x e A, then x is an element; hence

x e all. f

1.18 Definition If two classes have no elements in commor they are said to
be

disjoint. ln symbols,
A and B are disjoint

iff

AnB:@.

1.19 Definition The complement of a class ,4 is the class of all the elements
which do not belong to A. In symbols,

A'
Thus, x e A'

if and only if x (

{xl*d t\.

A.

Relations among classes can be represented graphically by means of a

useful device known as the Venn dagram. A class is represented by a simple
plane area (circular or oval in shape); if it is desired to show the complement
of a class, then the circle or oval is drawn within a rectangle which represents
the universal class. Thug .4 u B is rendered by the shaded area of Fig. 1,
A B by the shaded area of Fig.2, and A'by the shaded area of Fig. 3. The
reader will find'that Venn diagrams are helpful in guiding his reasoning about
classes, and that they give more meaning to set-theoretic formulas by making
them more concrete. For example, in Section 3 of this chapter we will prove
the formula

Ao(BuC):(AoB)v(AtC).
This formula is illustrated

in Fig. 4, where the shaded area represents

Classes and Sets

A o (B u C); one immediately notices that this same shaded area represents

(AnB)v(,4nC).

EXERCISES 1.2

c B and C c
a)(AvC)c(BuD),

D; prove that

1. Suppose that A

b)(.anOg(BnD).

Set 1.1.]
lHnt t IJse the result of Exercise 7, Exercise

2. Suppose

A: BandC: D;Provethat

a)AvC:BvD,
[Hnr:

b).4nC:BD.

Use the result of the preceding exercise.]

c B, then B' c A'.

of Exercise 4(a), Exercise Set 1.1.]
result
lHint,rJsethe
4. Prove rhat if A B, then A' B'.

3. Prove that If A

:
:
A
5. Prove thatif
6. Prove thatlf A
7. Prove Theorem
8. Let S

B
B and B
and

c
c

c CC,then A c C.

C,then A

1.11, parts (iv) and (v).

: {" | , I x}; use Russell's argument

to prove that S is not an element'

9. Does Axiom A2 allow us to form the "class of all classes"? Explain'

10. Explain why Russell's paradox and Berry's paradox cannot be produced by using
Axiom,A2.

THE ALGEBRA OF CLASSES

One of the most interesting and useful facts about classes is that under the
operations of union, intersection, and complementation they satisS certain

The Algebra of

Classes

31

algebraic laws from which we can. develop an algebra of classes. We shall see
later (Chapter 4) that the algebra of classes is merely one example of a dtructure
known as a Boolean algebra; another example is the "algebra of logic," where
v, A, r Lre regarded as operations on sentences.
Our purpose in this section is to develop the basic laws of the algebra of
classes. We remind the reader that the word clcss should be understood to
mean any collecton of objects; thug the laws we are about to present should
be thought ofas applying to every collection ofobjects; in particular, they apply

to all

sets.

1.20 Theorem If A and B are any classes, then

i\Ac.AvBandBc AvB,
ii)AaBcAandAaBcB.
Proof
i) To prove that A

c A'vB, we must show that x e A > x e A v B:

by1.5(i)
xeB
xeA+xeAv
+ xeAvB
by1.13.
c
A v B.
Analogously, we can show that B
ii) To prove that An B c,4, we must show that x e A B - x e A.

xeAaB + xeA n xeB

- ,r-{ I

by l.I4
by 1.s(ii).

1.21 Theorem If A and B are classes, then

A c. Bifand only if.4 v B

ii) A c Bifand onlyif A o B
1)

:
:

B,
A.

Proof

i) Letusfirstassume thatA c B;thatis,xel +xeB.

xeAvB-xeA or xeB
+xeB or xeB

+xeB
Thus,4 u B

c B;but B c

by 1.13
by 1.7(i)
by 1.8(iv).

A u Bby 1.20(i);consequently,Av B: B.
AvB: B. By 1.20(i), Ac AwB;

Conversely, let us assume that

thus .4 c B.

ii)

Then

The proof is left as an exercise for the reader.

32

Clsses and Sets

all classes
1.22 Theorem (Absorption Laws)' For

i)

ii) '4 n (AwB)-

Av(AaB): '

and B'

A'

Proof

i) By1.20(ii), AB c '4;therefore'bv1'21(i)'Av(AoB):
B):
ii) By 1.20(i),A = Au B;therefore'by 1'21(ii\'A (Av

A, (A')' :

1.23 Theorem For every class

Proof

'
A'l

'
by

xe(A',)',+x(A',+xeA

1.19,

by 1.19.1

xeA+xQA'+xe(A')'

all classes A and B'

1.24 Theore m (DeMorgan's Laws)' For

i)

(,a

u B)' : A' r B''

ii) (' n B)'

A'

B''

Proo.f

bY1'19

i)First, xe(AuB)'+x(AwB
+ xQ Aandx(

if either x e A otx e B' then x e A w B)

by 1.19
+ xe A'andxeB'
by r.14.
+xe (A'aB')
(because

Next, x e (A' n B)

ii)

by l.r4

i'x e A' andx e B'

- x( AandxfB
+x(AvB
>xe(AvB)'

The proof is left as an exercise for the eader'

by

1.19

by

1.19.

the following are true'

1.25 Theorem For all classes A, B and C'

commutatiue

Laws:

Idempotent Laws:

Assocatiae

r,aws:

i) A v B :

BwA

ii)AaB:BrA
iil\ Av A: A
lv)AoA':A
,ll

i:

::,"^:;"^z

[l : :]

The Algebra of Classes

Distributiue Laws:

vii) A n(B u C) : (A a B) v (A a C)
viii) Au (B n C) : (A v B) n (A w C)

Proqf

1)xeAvB+xeA or xeB
+xeB or xeA
+xeBwA
v) xe

Av(B

by

by 1.8(i)
by 1.13.
by 1.13

"':

by 1.13
by 1.8(ii)
by 1.13

l=t1ti:"i!.::l

> x(AwB)vC
vii) xe Aa(BuC) i xeA n xeBvC

ii.',ool 1: 1r",.,!'o ^

1.13

by

1.13.

by t.r4
by 1.13
xe

: lZl\u,,",i;:;,'

C)

by 1.8(iii)
by
by

l.I4
1.13.

The proofs of (ii), (iii), (iv), (vi), and (viii) are exercises for the reader.

The empty class and the universal class are identity elements for union
and intersection respectively; they satisfi the following simple rules:
1.26 Theorem For every class .4,

'l)Aw@:A.
iii) A v 4,/ * Q/.
v) '?l' :' g.
vii) I u A' : 4t..

ii)Aa:@.

iv) ,4 n 4t : A.
vi) Q' : a.
viii) ,4 n A' : @.

Proof
i) By t.I7, g A, and therefore by 1.21(i), A w @ + A.
iii) By 1.17(ii) A e al/, and therefore by 1.21(i), A v Ql : 4.
The proofs of the remaining parts of this theorem are left as an exercise for
the reader.

By using the laws of class algebra which we have developed above, we can
prove all the elementary properties of classes without referring to the definitions
of the symbols u, n, ', and
The following is an example of how such proofs
-c.
are carried out.

Classes and Sets

Example Prove that A a (A' u B)

A r B.

Proof

A n (A' u B)

: (A n A,) v (A a B)
: Ow(AnB)
:AB

by 1.25(vii)
by 1.26(viii)
by L26(i\.

The following definition is f'requently useful: The dffirence of two classes

class of all elements which belong to A, but do not belong to B.

A and B is the
In symbols,

A-B:AaB'.
Example provethat A

- B: B'-

A".

Proof

A-B:AaB'
:B'nA
: B' n (A,),
:B'-A'

Definition
by 1.2s(ii)
by t.23
Definition of B' A'.
It is useful to note that with the aid of rheorem 1.21, relations involving
inclusion (c), not merely equalit can be proved using class algebra.

EXERCTSES 1.3

1. Prove Theorem l.2l(1i).

2. Prove Theorem t.24(ii).

3. Prove Theorem 1.25, parrs (ii), (iii), (iv), (vi) and (vi).
4. Prove Theorem 1.26, parts (ii), (ii,), {v) through (viii).
5. Use class algebra to prove the following.
a) (A B) v C : (A v C) n (B v C), b) (,a u B) n C
6. Use class algebra to prove the following.

a\ If A C : A,then,4 n (B v C) : A n B.
b) lf A ^
a B : 6,thenz4 - B -- A.
-c) If A n B : and, Av B : C,then A : C
-

7. Using class algebra, prove ech of the following.

a)

Aa(B-C):(A^B)-C.
q - C : (A - c)u(B -

b) (.a v

C).

B.

(A

^C)

(B

n e.

Ordered Pairs Cartesian Products

c) A

d) A

u C) : (A - B) a(A - C).
n c) : (A - B) v (A - C\.
the operation + on classes as follows: lf A

35

(B
(B

8. We define

and B are classes, then

A+B:(A-B)v(B-A\.
Prove each of the following.

a\A+B:B+A,
, b\A+(B+C):(A+B)+c,
c) Aa(B-+ C):(AaB)+(A nC), d) A+ A: @, e) A+ @:
9.

A.

Prove each of the following.

ti)AvB:g+A:6andB:9.

b) A a B' : ifandonly tf A c B.
c) A + B -- gifandonly if A : B.
10. Prove each of the following.

a)

AwC:Bv

b) (,a u Q +

B)

11. Use class algebra to prove that

+BcC.

Cifandonlyif,4

(nu Q : @.+

if

C.

,4

c B and C : B-

r4, then

C.

ORDERED PAIRS CARTESIAN PRODUCTS

If a is an element, we may use the axiom of class construction to form the class

{a} : {x l*

o).

It is easy to see that {a} contains only one elemen! namely the element a.
A class containing a single element is called a singleton.
lf a and b are elementq we may use the axiom of class construction to
form the class
{o,b}

: {, lx: a or x:

b\.

clearly {a, b} contains two elementg namely the elements a and b. A class
containing exactly two elements is called an unordered pair, r, more simply,
a doubleton.

In like fashion, we can form the classes {a, b, cl, {a, b, c, d), and so on.
Frequently, in mathematics, we need to form ciasies whose elements are
doubletons. In order to be able to do tiris legitimately, we need a new axiom
which will guarantee that if a and, b are elements, then the doubleton
{a, b} is
an element. This motivates our next axiom, which is often called the Axiom
of Pairing:

A3. If a and b are elements , then {a, } is an element.

It is clear fhat {a, a}': la}; thus, setting a : b in Axiom A3, we immedi.
ately get

if

is an element, then the singleton {a} is an element..il

36

Classes and Sets

1.27 Theorem

If {x, y}

{u, u}, then

lx: u and y: uf

or

lx: u and y:

The proof is left as an exercise for the reader.

!, separately. Use Axiom Al.]

lHint: consider the

u].
cases

y,

An important notion in mathematics is that of an ordereil pcrr of elements.

Intuitively, an ordered pair is a class consisting of two elements in a specified
order. In fac the order is not really essential; what is essential is that ordered
pp{rs have the following property.
t

1.28 Let (a,b) and (c,d)

b:d.

ordered

pairs. If {a,b): (c,d\ then a : c and

we would like to define ordered pain in such a way as to. avoid introducing
a new undfined notion of "order." It is an interesting fact that this canl
indeed, be accomplished;we proceed as follows.

129 Definition Let

be the class

a and,b be elements; the ordered pair (a,) is defined to

(a,b)

{{a}, {a,b}}.

By Axiom A3, (a, ) can be legitimately formed, and is tserf an erement.

It is worth noting that

(b,

a\

{{b}, {b, o}}

{{b}, {a,b}}.

Hence there is a clear distinction between the two possible ..orders,, (a,
b) and
(b, a): they are distinct classes. It remains to provi that ordered
pairg as we
have just defined therr have property 1.2g.

1.30 Theorem If (a, b):

Proof.

Suppose that

(c,d),thena: candb:

(a,bl

(c,

d); that

{{o}, {o, b} }

d.

is,

: {{c}, {c, d}}.

By Theorem 1.27. either

[{r} : {"} and {a,b} : {r,it}],

Ordered Pairs Cartesian Products

or

l{a} : {c,d} and {a,b} : {c}l;

we will consider these two cases separately.

Casel. {o} : {"} and{a,b} : {",d}. From {o\ : {r\,itfollows thata: c.

From {a, b\ : {",d} and Theorem 1.21, it follows that either a: c and
b : d, or a : d and b : c; :r'the first casg we are done; in the second case,
we have b

d, so again we are done.

2. {o): {t,d} and {a,b} : {c}. Here c e{c,d\ and {c,d} :

{a}; thus c : q; analogously, d = a. Also, b e {a,b} and {o,b} :
b e {c\; hence b : c. Thus : b : c : d,and wearedone. I
Case

ce

{a},
{c},

so
so

1.31 Definition The Cartesian proiluct of two classes A and B is the class of
all ordered pairs (x, y) where x e A and y e B. ln symbols,
'

A x.B: {(*. illxe l

and yeBI.

The followin g zre a few simple properties of Cartesian products.

1.32 Theorem For all classes A, B.and, C,
i\ A x (B n C) : (A x B\ a(A x C).

A x (B v C) : (A x
1i1) (A x B) n (C x D) :
1l)

u(,4 x C).
(A o C) x (B n D).

B)

Proqf

1)(x,y)eAx (BnC)+xeA and yeBnC

+xeA and yeB and yeC
o(x,y)eAxB and (x,y)e AxC
iii)

(x, y) e

(A x

o(x,y)e(AxB)n(AxC).
B)o(C xD)e(x,y)eAxB and (x,y)eCxD
+xeA and yeB and xeC and yeD
xeArC and yeBaD
.+ (x, y) e (A a C) x (B n D). I
it

instructive to represent relations between classes by

it is often convenient to illustrate relations between
products of classes by using a graphic device known as a coordinate diagram.
A coordinate diagram is analogous to the familiar Cartesian coordinate plane;
there are two axes-a vertical one and a horizontal one-but we consider only
Just as we found

means of Venn diagramg

Classes and Sets

one "quadrant." If we wish to represent a class A x B, then a segment of the

horizontal axis is marked off to represent A and a segment of the vertical axis
is marked off to represent B; A x B is the rectaigle determined by these two
segments (Fig. 5). As an example of the use of coordinate diagrams, Theorem
1.32(iit) is illustrated in Fig. 6.

"J

BD

rI

"I

I
I

ncl

Fig. 5

EXERCISES 1.4
3r!

t.Let A:{",b,c,d}, B:{r,2,3}, C:{r,y,"}. Find,4 xB, BxA, Cx(Bx

(A v B) x C, (A x C) ur (B x C), (Aw B) x (B u C).
2. Prove Theorem 1.32(ii).
3. Prove that'.4 x (B - Dl
4. Prove tbat(A x
5. lf A, B
a) (A

and,

(A x Bl - (A x D).
B) n(C x D) : (A x D) n (C x

B).

C are classes, prove the following.

x A) a(B x

b) (,a x D - (c
c) (A x A) - (B

C)

(A a B\

x C) : l@ x C): f(A -

x (A

C\.

C) x Bl u la x (n - Ql.
B) x AlwlA x (A - C)f.

6. Prove that A and B are disjoint if and only if, for any nonempty class C, A
B x Caredisjoint.

7.lf AandCarenonemptyolasses,provethat A c
BxD.

BandC

8. LetA.B,C,Dbenonemptyclasses. Provethat A x

andB:D.

9. lf A, B, and C are any

?) ,

.ir.

plasses,

B:

ifandonlyif,4

Difandonly if

prove

* BandA'x Caredisjoint, b) B x AandC x

A x B : ifandonly if A : A or B : @.

10. Prove that

A),

A'aredisjoint.

C and

xCc

A:

Graphs

11,

39

Prove each of the following.

a) lf a: {b},tbenb e a.
b) x: yif andonlyif {x}: {y}.
c) x e a if and only 1f {x\ c q.

d) {a,b}

{a} ifand only if a

b.

12. We give the following alternative definition of ordered pairs:

(x,y):{{,,@},{t,{}}}uu"nr,}Tl";rl,T;!,il_7."

GRAPHS

A class of ordered pairs is called a graph. In other words, a graph is an arbitrary

subclass of all xi?/ .
The importance of graphs will become apparent to the reader in Chapters
2 and 3. It may be shown, for instance, that a function from A to B is a graph
G c A x B with certain .special properties. Specifically, G consists of all the
pairs (x, y) such that y : /(x). This example may help to motivate the following
definitions.

1.33

.Definition

1.34 Definition
follows:

If G is a grapt\ then G-l is the graph defined by

G-t : {(r, y) I $, x) e G}.

If

G ,and

G. H

{$r:

H arc grapfus, then G ".EI is the graph defined

{(r,y) I 1z=(x, z)e

as

H and (z,y)e G\.

The followin g are a few basic properties of graphs.

1.35 Theorem

If G, 1, and J

are graphs, then the following statements hold:

i) (G. H)"J: G.(II.J):

ii) 1C-t-t * O.
iii) (G " l)-t : H-L " G-r.
Proqf
i) (x' v) e (G ' r{) "

It
j
I

:i1= *u''=:: r"::i. ,o;i]^'ri,l. H

' +lw=(x,w)eH.J
(x,y)e6"(H"J).

and (w,y)eG

and (w, y) e G

Classes and Sets

ii)

(x, y) e

(c-1)-t o (y, x) e G-L

+(x,y)eG.

iii) (x'Y)e (G

'H)-':T1'#1i1;:ru
"',izu '

1.36 Definition Let G be a graph. By the ilomqin of G we mean the class

domG: {xl3ya(x,y)eG},
and by the range of G we mean the class

ranG+{yll*=(x,y)eG}.
ln other words, the domain of G is the class of all "first
elements of G, and the range of G is the class of
elements of G.

components" of
all "second components" of

1.37'Theorem IfG and Il are graphs, then

i) dom G : ran G-t,
ii) ran G : dom G-1,
iii) dom (G " H) c dom 1, iv) ran (G " H) c ran G.

Proqf

i) xedom G+1y=(x,y)eG
e3y= (y,x)eG-1

xcranGiii) xedom(G

3y:(x. y)e(G"H\
+12=(x.z)e H and (z,y)e G
+_\domFI. I

"H)-

1.38 Corollary Let G and Il be graphs. If ran .EI c dom G then dom G o H
dom H.
The proof of this theorem is left as an exercise for the reader.

EXERCTSES 1.5

l.

Let

{(b, b), (b, c), (c" c), (c, d)}

Graphs

and

Find G-1,

H : {A.a\(c.bl,(d.c)}.
H-',G " H,H " G,(c " H)-t,(G v H)-1,H-t

2. Prove Theorem

1.37, parts

3. Prove Theorem

1.38.

4. If

G,

U, andJ

" G.

(ii) and (iv).

are graphs, prove each of the following.

(Hu J)"G: (H.G)u(J"G),

c) G"(H n,r) c (c"H)n(G""/),

a)

b) (C

- Hl-r:G-l - H-t,
-(c"J)=G"(H -

d) (c" H)

5. lf G and H are graphs. prove each of the following.

a) (G n Hl-t : G-1 H'1,
b) (G u H)-'

J).

: G-t v H*1.

6. lf G, H, J , and K are graphs, prove

a) ifG c H andJc K,then G " J c H " K,
b) C c Hlf andonlyif G-1 s H-t.

7.IfA,B,andCareclasses,proveeachofthefollowing.

a)(AxB)-':BxA.
b) If,4
O,then(,4 x B)"(A x B): A x B.
^B B+are
c) ll,4 and
disjoint, then (,4 x B) " (A x q : A.
d) If B / @,then(B x C) "(A x B\ -- A x C.
8. Let G and H

be

graphs; prove each ofthe following.

a) IfG c A x B,thenG-1 c B x A.
b)IfG-AxBandIIcBxC,thenIoGcAxC.

t&

9. If G and H are graphs, prove each of the following.

a) dom (G

v H):

(dom G)

(dom H).

b) ran (G u H) : (ran G) u (ran H).

c) domG - dom H c dom (G - H).
d) ran G - ran H c ran (G Il).

10. Let G be a graph, and let B be a subclass ol the domain of

G to B we mean the graph

Gror:

{(x,

G. By fhe restriction of '

y)l@,y)eGand xeB}.

Prove each of the following.

b) Ga,cr : Ga1v Gc1,

d) (G. H)s: G " Htut.

a) Gn: G o(B x ranG),

s) Ga^cl : G.\ | Grq,

11. Let G be a graph and let B be a subclass of the domain of

G(B) to designate the class
G(B)
Prove each of the following.

a) G(B) : ran Gr,

c) G(B n C) : G(B) a

{y

l,

eB

G.

We use the symbol

(x, y) e G\.

b) G(B u C) : c(B) u G(C),

d) If B c C, then c(B) s c(C).
"

G(C),

Classes and Sets

GENERALIZED UNION AND INTERSECTION

A,\; its elements

are indexed by the numbers

1,2, ..., n. Such a class il often calld an indexed family of classes; the numbers
1,2,...,n are called indices and the class {1,2,...,n} is called the index class.
More generally, we are frequently led to think of a class I whose elements
i,j,k,... serve as indices to designate the elements of a class {A,Aj,Au,...}.
The class {Ar, A, Ao, ...} is called at indexed family of classes, 1 is called its
inilex class, and ihe elements of 1 are called indices. A compact notation which
is often used to designate the class {A, A,,40, ...} is

Consider the class {Ap A2,...,

{At}or'
Thus, speaking informally, {A,\n, is the class of all the classes A, as i ranges

/.

over

Remark. The definition of an indexed family of classes which we have just

given is, admittedl an intuitive one;it relies on the intuitive notion of indexing.
ihis intuitive definition is adequate at the present time; however, fr future
reference, we now give a.formal definition of the same concept:
By an indexed family of classeg {,A,},.r, we mean a graph G whose domain
is 1; for each e / we define ,4, bY

A,:

{x l1;,xec}.
- {1,2\, A,.: {a,b}, and

where.I

For*grample, consider {A}n

Theff, formally, {Atli.r is the graph

Ar:

{c,d\-

{(t, a), (1, b), (2, c), (2, d)}.

If {Ai}rd is an indexed family of classes

such that for each

iel, A,is'an

element, then we let {A,l ; e f} designate the class whose elements are all the
A,, that.is, {,4, | ; e f} : {t l, : .4, for some i e I}. However, we shall follow
current mathernatical usage and use the two expressions, {,4,},.. and {Arl ; e f},
interchangeably.

1.39 Definition Let {A,},., be an indexed family of classes. The union of the
classes, , consists of all the elements which belong to at least one class .4, of
the family. In symbols,

UA,:

{xll.e I=xeA1.

The intersection of the classes,4., consists of all the elements which belong to
every class A, of the family. In symbols,

)A,:
ieI

{xlv;e I,xeA,\.

The following are some basic properties of indexed families of classes.

ho - t vv

a 0 s- u"c t''A e"

tato \".tt'-t-.
/vo ob:k a k

Lo

vl,r"o

o,

, totr,oo,

\l- n,\uch: f ut \ff\tr!, s Q'^- vYrvc\. h.*

lntersection

43

xe[J A,, than xeA,

for

Generalized Union and

1.40 Theorem Let {A,},., be an indexed family of classes.

i) If At c B for everY i e 1, then fl '1, n.

ii) IfB

Aforevery i

el,

then U

- )O,.

Proqf

i)

Suppose that

some

A,c B for every iel;

j e 1;but Aj -

B, so x e

now if

B. Thus [J A,
=
ieI

B.

The proof of (ii) is left as an exercise for the reader.

iel

1.41 Theorem (Generalized deMorgan's Laws). Let {A,\n, be an index family

ofclasses. Then,

: ),q;.
iel
ii) (l^) .4,)' :0 ,q;:
i) (U /,)'
iel

iel

ieI

Proo.f

i) x e (lJ A,)' o xdU

iel

A,

iel

+VjeI,x(A,
YieI,xeAr'

+xe

,]ot''

The proof of (ii) is left as an exercise for the reader.

1.42 Theorem (Generalized Distributiue
indexed families of classes. Then

i) (U
ii)

,4,)
i.4
(n ,4,)
iel

Laws). Let lA,j,., and {Br}r.,

n ([J B.) :

U (A, n B),
u (l B;) : 0 (A, v B).
jeJ
(i,j)el J
jeJ

(i,j\el

xJ

Proof

i) xe([J,4,) n(U
ieI

be

B,)+x.U A, and xel) n,

<> x ,to ,o. ,o-" n. ll x e Bofor some k e J
<>x Ao o Bofor some (h, k) e I x J
<>x U (A,aB).
. (i,i)eI J
x

Classes and Sets.

The proof of (ii) is left as an exercise for the reader.

A theorem concerning the union of graphs will be useful to us in the next

chapter.

1.43 Theorem Let {G,},.t be a family of graphs. Then

i) dom (U c,) : ! (dom Gr).
iel

ieI

ii) ran (U C,) : [J (ran

Gr).

iI

iel

Proqf

xedom(U G,)e3ya(",y)e U

i)

t"

bY 1.36

+ 3y : (x, y) e ]ro, ,o.n.;. r

+ x e dom G, for some j e f
+ x lJ (dom G')

by

1.39

bY 1.36
bY 1'39-

ieI

The proof of (ii) is left as an exercise for the reader.

A variant notation for the union and intersection of a family of classes

is

sometimes useful. If . is a class (its elements are necessarily classes). we define

the union o.f ., or unon of the elements o.f , to be the union of all the classes
which are elements of ,. In sYmbols,

l) A: {xlxe ,4.forsome Ae.\.

In other words, *.^" oif and only if there is a class 1 such that x e A
1.4

and

Ae'&

Analogousl we define the intersection of d, ot intetsection of the

elements of ., to be the intersection of all the classes which are elements of
..Insymbols,

Ae,.

) e:

1.45

{xlxe.4forevery Ae.}.

Ae.r'/

1.46 .Example Let

u: {d,e}. Then

,d

{K,L, M},

where

{s,b,d\,

) A:
U A: {a,b,c,d,e\ and Ae"t

f : {a,c.it\, and
{d\.

't.

1.47 Remark

It is frequent practice, in the literature of set theory, to write

Ud for l)

t/

Generalized Union and

lntersection

45

and

for ),.

)d

Ae,ry'

We shall occasionally follow that practice in this book.

EXERCISES 1.6

1. Prove Theorem 1.40(ii).

2. Prove Theorem 1.41(ii).

3. Prove Theorem

1.42(11\.

4. Prove Theorem 1.43(ii). ,

5. Let {A,},,, and {8,},., be two families of
that Vi e I, A, c Br; prove that

with the same index class

I. Suppose

b)0.a,sn4.

a)UA,cU4,
eI

classes

eI

eI

iel

6. Let {A,}r., and {^B,}n, be indexed families of classes. Prove the following.
(A, x B1,
a) (n.4,) x
B;)

(ljeJ : (i,j\eln J
b) (U,4,) x ([J B.) : U (A, x B).
jeJ
(,j\el J
eI
ie/

7. Let {A,}r., and {B}r be indexed families of classes. Suppose that V

B,

4,.

eI,)jeJ=

Prove that

)a, s ).a,.

8. Let {A,}r.,and {B};..r be indexed families of classes. Prove that

(n lA,- Bl),
a) (U 1,) (U )
U
jeJ
iel
iel jel
b) (n.4,) (llBj) n (u lA,
Bf).
iel

jeJ

:
:

eI je,I

Aif A
Suppose that
= !J 4.
{.B,}". and {C,}ru are two distinct coverings of ,4. Prove that the family
{(4 ^ c)}rt,.r*, is a covering of .4.
10. Let a: {u,u,w},b: {r,r}, c: {w,y},r: {a,b\,s-- {b,"},andp: {r,s}. Find
9. We say that an indexed family {8,},., isa couerng of

u(np) n(up).
Prove that a(. v g) : (a.) n (n0).
the classes u(up), n(np),

11.

12. Prove each ofthe following.

a\ If A e 9,then A c v Q and a0 c A.
b) U c 0if andonlyif w. c vQ.
c) lf @ e,,then a, : .

16

Classes and Sets

SETS

undoubtedly, everything we have said in the preceding pages is fairly familiar

to the readei. nu* tftorrgh we said,class where the reader is more accustomed
in
io.hearing sef, it is obvius that'the "union" and "intersection" defined
the
sets,
of
intersection
and
this chapter are prcisely the familiar union
..Cartesn product" is eiactly the usual Cartesiannproduct of sets, and simi'At this poing it appears
larly for the other concepts introduced in this chapter.
sets can be done with
with
to
doing
as trough everything *" ur" accustomed
bother distinguishing
"why
aslg
classes. Thus, the reader may very well
in terms of
mathematics
of
all
develop
between classes and sets? Why noi
elements,"
of
collection
"anf
means
classes?,, Since, as we have said, a class
to this
answer
The
it?
with
done
be
and
why not simply call a class a se!
explain
is
to
section
this
of
purpose
chief
l,rstiol is oi lreat importance; the
*hy t"" do'want to distinguish between classes and sets'
First, we note that the axiom of class construction (Axiom A2) permits us
to form the class of all elemens which satisfy a given property; it does not
allow us to form the class of all classes which satis$ a given property' The
the class of
reason for this limitation is obvious: if we were enabled to form
"Russell's
form
could_
we
then
property,
givan
all classes which satisff any
give
would
this
and
of
themselveg
elements
not
class" of all classes which aie
us Russell's paradox.

Next, we note that in mathematics we often need to form particular

of sets. A few examples which come to mind are the following:

ses

The set of all closed intervals la, b] of real numbers'

The set of all convergent sequences of real numbers'
points)'
The set of all the lines in the plane (where each line is regarded as a set of

Let us look more closely at the first example; in a discussion in elementary

calculus, we would feel perfectly free to say "let .-ql consist of all the sets which
are closld intervals la,-b] of real numbers." Now if "sets" were no different
from "classes" then, by the preceding paragraph, we would not be allowed to
form.. This would b an iniolerable restriction upon our freedom of operating
with sets in mathematics.
Let us recapitulate: The notion of class is appealing because of its irituitive
simplicity and lenerality; however, there is a serious drawback to dealing with
nu-.t/ttrut it is not permissible [for an arbitrary property P(X)] to
"lur."r,
form tie "class of all classes X which satisff P(X)." This would be an intolerable restriction on our mathematical freedom of action if we were to base
mathematics upon classes. Instead we base mathematics upon ses; the
concept of a sef is somewhat narrower than that of a class; sets will be defin"d

Sets

47

i"\, svch;a
JfLy; h-o_wg.u.el, that .for any propertv p(X\, it is legitimate to form the
class o.l atl sets
which satisy
Thus, the freedom we require is resioFd

p(x).

I with sets rather than the broader notion

of classes. we are more than willing to do so, because, as we will be able to
show, all the classes we deal with in mathematics are sets.
Ig\!-b.ow Should sets be defined? In order to answ"' this-que"ti.rn, it is
e_ssential to remind ourselves, once again, fht we are seeking a way of defining
t
n element. Then
it us to
all sefs v which satisfv anv prooerty p(X).
This simple answer is, in fact, the one which has been adopted by most

same thin
form the class of

mathematicians. We will use

1.48 Definition By a set

is.

it here; thus,

meant any class which is an element of a class.

ihir d.finition is supported by our intuitive perception of what a set should

be. For if A, B, c, ... are'sets, it is perfectly reasonable that we should be able
to form the class .{A, B, C, ...} whose elements are A, B, C, .... In other

words, we would quite certainly expect every set to be an element. The converse
is equally reasonable: for i[.] !s not a set, then A is a proper class. and we have
atready seen that ln order to avoid contradictions, proper classes should not be
elements of anything. Thus, if I is not a set, then I is not an element.

In the remainder of this section, we will state the basic axioms dealing with
sets. The main purpose of these axioms is to guarantee that when the usual
set-theoretic operations are perform'ed on sets, the result, each time, is a set.
First, we note that the Axiom of Pairing, our Axiom 43, may be re-stated

thus:

43.. If a and b are

sets, then {a, } is a set.

Now, if I is a set and A

,8, one would reasonably expect I to be a set.
This is the content of our next=axiom, called the Axiom of sibsets.

44.
A

Every subclass of a set is a set.

By Theorem 1.20(ii), A n B c l; thus, by Axiom A4,if A is a set, then

B is a set. In particular, the intersection of any two sets is a set.
The union of "not too many" sets should be a set. This is guaranteed by

our next axiom, called the Axiom of Unions:

A5. If . is a set of sets,

then

Aed

is a set.

rf A and -B are sets, then, by Axiom A3, {A, B} is a set; it follows immediately

48

Classes and Sts

l-l X : A u .B; thus, by Axiom A5, Au B is a

set. This shows that ,n, ,fi'] two sets is a set.
from Definition 1.44 that

1.49 Remark. By Axiom A3,. every doubleton is a set. Furthermore, letting

a * b in A3, every singleton is a set. Since the union of two sets is a set,
follows that every class of three elements is a set, every class of four elements
is a set, and so on. Thus, in an intuitive sense, every finite class is a set.

Next, we will establish that if is a set, then the class of all the subsets
is a set. We begin with a definition.

of

1.50 Definition Let Abeaset; bythepowersetof Awemeantheclassof all

A. In symbols, thepower set of I is the class

the subsets af

g(A): {BlB q A}.

Note that by Artiom A4, g(A) is the class of all the ses ,B which satisfy
By Definition 1.48 and Axiom A2, it is legitimate to form this class.

A.

The following is called the Axiom of power Sets:

A6. If

is a set, then

1.51 Example

If A

g(A) is a set.
,o,], then

g(A)

: {, {a}, {b}, {a, b}}.

From all that we have said so far, it does not yet follow that there exisr
all. To fill this vacuum, we state a temporary axiom, which will be
superseded by Axiom A9:
any sets at

The empty class is a set.

Henceforth, we will refer to @ as the empty sel.

From Axiom T, together with 43 and 45, we may infer the existence of
a great many sets. We have the empty set, @; we have singletons such as
{}, {{A}}, etc.; we have doubletons such es {, {@}}, forried by any two
of the above. Similarly, taking unions of the above'repeatedly, we may form
sets with any

fnite

number of elements.

1.52 Remark An important consequence of Axiom 46 is the following.

is a set, then clearly

lf

,4

B:{Xlxet
and p(x)}
is the class of all the subsets of r which satisfy the property p; by Axiom
44, Axiom A2 may legitimately be used to form ttre class .8. Now if x e B,

Sets

49

then Xis a subset of

so X e 9(A); thus .B c 9(A), But by Axiom A6, g(A)
is a set, hence by Axiom A4, .B is a set, We may summarize as follows: if is a
set and P(X) is a property of X, then the class of all the subsets of A which satisfy
P(Xl is a set.

l,

A andB are setg then I x B is a set.

Proof. Let A and B be sets. By Axiom A5, AvB is a set; by Axiom A6,
9(A v B) is a set; finally, by Axiom A6 again, glg(A u B)] is a set. We will
prove that A x B glg(A u B)], and it will follow, by Axiom ,A.4, that
=
A x Bisaset.
Let(x, y)e A x B. By 1.29,(x,y): {{r}, {r,y}}. Now xeAv B,hence
{*} = 'u B, so {x} eA(Aw B). Similarl xe AvB and ye A w B, so
{*, y} s A v B, hence {x, y} e 9(A w B). We have just shown that {x} and
{x, y} ae elements of 9(A u B), hence
{ {"}, {r, y} } s 9(A v B);

1.53 Theorem lf

it follows that

that is,

{r}, {r, y}} e ele(A u

(x,

Thus

"y)

qlgl

B)1,

B)1.

AxBcglg(AuB)1.

It follows from Theorem 1.53 arrd Axiom A4 that if A and B are sets, then
anygraph G c A x Bisaset.
It is easy to show that if G is a set, then dom G and ran G are sets (see
Exercise 5, Exercise Set 1.7). Using this fact, one can easily show that if G and
El are sets, then G " H and G-r are sets (see Exercise 6, Exercise Set 1.7).
EXERCTSES 1.7

1. rf

u4

and B are sets, prove that A

Set 1.3.)

-B

ado

ui,firP,i?;llse

8, Exercise

class and A c B, prove that B is a proper class. Conclude that the

union of two proper classes is a proper class.
tla*
^.r
3. Prove that the "Russell class" and the univrsal class are proper classes. [Ainr$se
the result of Exercise 8, Exercise Set 1.2.]

2. lf A is a proper

4. Let {Ail.r

be an indexed family of sets. Prove

that (-)

,4, is a set.

a graph. Prove that if G is a set, then o- 'rrr ran G are sets. [Ifinr: Show
that both dom G and ran G are subsets of u(uG).]

5. I,et G be

50

Classs and Sets

G- and G o -EI are sets'

6. ].t G and II be graphs. Prove that if G and f[ are sets, then
and 9(vp)'
7. Let r : {o, b}, s : {b, c}, p : {r, s}. Find the sets 0(r\, 9(9(r))'

8. Let Aand B be sets; prove the following'

' a) A B if and onlY 1f 9(A) = E(B)'
=
b) A: B if and olY 1f 9(A) = 9(B\'
c) E(A) 9(B) = e(A B)'
d) E(A) w 9(B) c 0(A v B)'
t' n' n : gJ 1f arrd ov tf 9(A) 9(B)

"\

9. lf A and I arc sets,prove the following'

a\ v(e(6\): fi.
c) lf 9(A) e A(0\
10. Exhibit the sets

then A e

s(s())

: {q}'

0 n(s(s\): '

A'

and

Els(E(@\\l'