11 .
On the Method of determining , from the real Probabilities of Lfe , the Values of Contingent fReversions in which three Lives are involved in the Survivorship .
By William Morgan , Esq. F.IRS .
Read December 12 , 1799 .
ON SURVIVORSHIPS .
rfHE several papers which I have had the honour of communicating to the Royal Society , on the doctrine of contingent reversions , contain the greater number of those cases in which three lives are concerned in the survivorship .
With the view of completing this subject , I have been induced to investigate the remaining problems ; and , having succeeded in the solution of them , I hope the following will not be considered as an improper addition to my former communications .
Being anxious to render this paper as concise as possible , I have omitted to state at length the different contingencies on which the payment of the given sum depends ; trusting that , from an attentive perusal --of my former demonstrations , these will appear to be so plainly expressed by the several fractions in each problem , as to render a more ample description of them unnecessary .
PROBLEM I. To determine the value of a given sum , payable on the death of A or B , should either of them be the first or second that fails , of the three lives , A , B. and C. Solution .
In this case , the payment of the given sum must certainly take place on the extinction of the joint lives of A and B , independent of C , and therefore the value of the reversion will be S. r _ .
VAB r The fractions expressing the contingencies on which the payment of S depends , in the ist year , are abr x a ' .
b -ln.cd + a ' .
- .
d +a.c-d.m + b. c d. aa S5I acr x abc-mc .
a-a ' , in the 2d year-gr xa m-n , de + a " .
m m-n .
e+ all -x de .
n+ all n Xe + m-n .
d e.a a ' +a"+m-naa'+a".e+cd.n.a " -it 51 C c-d .
m -n a_ a ' + a"l + cd .
m-n.all -ac , x mc .
aan c. a a ' + a"l , and so on in the other years ; hence the whole value is xr -1 .
V-r .
AB + AB1 Sr-I V -AB , as before .
Q. E. D. PROBLEM II .
To determine the value of a given sum , payable on the decease of A or B , should either of -them be the second or third that shall fail , of the three lives , A , B , and C. * The same symbols are uniformly retained in this , as in my last two papers on the subject .
See Phil. Trans. Vol. LXXXI .
page z47 .
Solution .
In the 1st year , the value of the reversion will be a b-r into r~ a.c-d .
m b-mxed.a-a ' a.crabm dr+ '~~ in the 2d year it will be a bc r " into a".m n.d -e + a " .
m-n al. d -e nm it .
d e.a_ + al t+ all e2++ a,.i e+bmd -e .
a"+ c -d .
m -n .
a"+ a ' , m -n.e 4b i.a".e +c-d.m-n.a a'+a"+c-d.n.aII+acdm b nt- ?
_d *aD ; in the 3d year it will be b intoa nlO _~~~ ~~~~~ af of ee a ' all of + a " .
no f4+ n-o .
e-f .
Y~'4 .
n _+ b.n. _aa c-e.n-o.f __ 2 +n-o.f.a'+aa " + b-n.f.a"'r+c-e.n-oaa'+a"l+a " ' ca"'4 .
c -B e. n -o .
d+ a " b. c e. a " ' +c e. o* .
all+ -+ zand so on .
These different fractions being added together , will be found S b-m mn noS b-m.c m-n .
d=x+ t* r ; _ + r3 ) @zb X+ r ; brr ~~~r3 'rr n-o~e Sb m. d in n.e n-o.f +3 &amp; .
--x + 1+ C+ -3 &amp; C.6rb Xrr r3 rS C* 4 .
S at S Ca da " 77 x. 7++p+ , &amp; C.+73X .
+ .
I+ , &amp;c. Sb a ' m al.i a a " ' Sbc a ' mnda ' +F++ rXJ 3+ SabcXr fr , + 1 ?
, e z1a ` ox S in ca ' nd a ' o a " ' S3+ &amp; C- &lt; * b x+ SC .
+ &amp; C r2 ~~~a bcr r2.2a mdd nea of a " ' Sma n. a'+ all xr+ i ' + r3 &amp; +C .
+ xab + o0a +'a " ?
a " ' S ba ' m. a al S r3 &amp; C. ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~-----+ -.-+ &amp; C. ab rr zabc bc a ' md.a+ all .
Smc a ' nd a+ a " X r-~ -r &amp; zaC cX r +-rz &amp; C* Sb da ' me a'+ a " S md .
ad + TzaJX + , &amp; C. b ra +c &amp; r + 77X +zab X-r + -= -+ , r ' + SC .
J~~~ abr Xr +r+2 aC bX nl .
a ' oa( ?
&amp; t. S md 'd no a(+aft C. + abcrX r r~~~~~..r , 2 a. cSnd a ' o a ' + Sd-'c 4 .
&amp;c. which may at last be re2absr rr duced to S into lx V+ ABC _ s.j-c AC r 2 : 2r X~~~~~~ AB 2C XA + BK9 , ABK +2b X.4 AFC +bx1+ AP + d.PT-APT But , unless A and B are very nearly of the same age , and both older than C , this rule will not be sufficiently accurate .
If B be the oldest of the three lives , the annuities A , AC , and AK , should be continued only for as many years ( x ) as are equal to the difference between the age of B and that of the oldest life in the table of observations .
Let those annuities be respectively denoted by A ' , A'C ' , and A'K ' ; also let p denote the probability that C survives B , * q the number of persons living opposite to the age of A at the end of x years , then will the value of S , after x years , be = tq X r-x , Vand the whole value of the reversion will be= =S into -x V+ ABC A+ +BCx AC AB r2zr Tx AIK+BK 92BK + xA F"AhJc+ _jxI+ AP d. PT-APT qr_ X. V-Ax +c~+TXr ( Ax denoting the value of an annuity on a life x years older than A ) .
If C be the oldest of the three lives , let as , s -t , t -u &amp;c. be substituted for their equals a1 , a " , a " ' , &amp;c. and c ' , r -c'/ + C/ / C C ' + c " + c " ' , &amp;c. for their equals of , g. &amp;c. then will the value of the reversion , by pursuing the same steps as in the former case , be found =S into rX V-.--~A ?
B ?
ABC+BC~ AB+C 6 .
AF AFC+ v -L2A B +.ABC + --O AB +b a. HB-HBC s rn s 2C1 ba Xl1 + NC2-rX I+ PC +bX+ NPC .
But , as these series are to be continued only during C 's life , it is evident that the annuities A , B , AB , &amp;c. should also be Continued only during this term ; and therefore , if z be the difference between the age of C and the oldest person in the Table , A ' , B ' , A'B ' , &amp;c. the values of annuities respectively on the single and joint lives of A , B , A and B , &amp;c. for z years , these several symbols should be substituted above , in lieu of A , B , AB , &amp;c. to denote the value of the reversion during the first z years .
After the extinction of the life of C , the given sum may be received upon either of three events ; ist , if A should have died before C , and B died after him in the z+ 1 ' Z+ 2 , &amp;c. years 2dly , if B should have died before C , and A died after him in those years respectively ; 3dly , if both the lives of A and B should die after the first z years .
Let v denote the probability that C dies after A , and wr the probability that C dies after B , * then will the value of the reversion depending on these several contingencies ( putting p for the number of persons living opposite the age of B at the end of z years , and q for the same number opposite the age of A ) be =Sinto p* tr-I .
VBx qr .
r-IV_ .
%p Pq I+ABP into h a+ ?
rz+ + abrx Xr ABz , and the whole value of the reversion will be S into Al+ B r.V_-+ ABC BCA ' ' 2v+ ?
BC+AC AB + ( .
A'FAFC+ .
HB'_HBC rzr2baaS1+ NC+. .
~x SL .
I+PNC I+PC p. ( rI .
V -13 zarX -b ra2 r_ _ -+ -b rR + _q VA + " P q:+ x1+ AB~Zf If the lives be all equal , the value , according to the first rule , will beS into rtV + CC CC + CC cc 5 , 1 ~~r -2+ ( r , d dd rd-0-+-yxKCC --KC + -6xi +C ?
JT+ zccr XTTT-CTT ; and , according to the ?
d rule , it will be-S intor , xr --t ra &gt; &gt; v Cc K-CCK d\{ _\ , _ _C cc , \ _ + , _ XI+ CT + dd x'i + CTT .
If these expressions be resolved into their respective series , the value in each case will be found SsrxVC CC + CCC , which is known to be the true value , from self-evident principles .
But the solution of this problem may be obtained by the assistance of the ist problem in my last paper , * supposing , instead of a lgyen sum , it were required to know the value of the reversion of a given estate .
For , since the possession of this estate is an event which must certainly take place , and the only point to be determined is the time in which it will probably happen , it is obvious that no event can postpone the possession , but the contingency of C 's being the second that fails , of the three lives .
If , therefore , the sum of t'ie valucs of an annuity on the life of B after A , provided A should die before C , and of an annuity on the life of A after B , provided B should die before C , ( both found by the problem just mentioned , ) be subtracted from the whole value of the reversion after the joint lives of A and B , the remainder will be the value required .
Let X and Y respectively denote the annuities found by problem 1St , ( Phil. Trans. Vol. LXXXIV .
) then will the general rule expressing the value of an estate be =V AB X+ Y , and consequently of a given sum -S rxV AB X+ Y , which , when the lives are'equal , may be reduced , as in the former cases , to S r-i xV-(C--CC + CCC .
Q. E. D. PROBLEM III .
To determine the value of an estate , or of a given sum , after the decease of A or B , should either of them be thefirst or last that shall fail , of the three lives , A , B , and C. Solution .
The reversion of the estate in this problem , like that in the preceding one , cannot be prevented ultimately from taking place ; and there is only the single contingency of C 's being the first t~iat fails , of the three lives , which can postpone the possession of it after the extinction of the joint lives .
The whole value , therefore , of the reversion , after the joint lives of A and B , must in this case be lessened by the sum of the values of an annuity on A 's life after B , provided B should survive C ; and of an annuity on B 's life after A , provided A should survive C , ( both found by the 2d problem in my last Paper.* ) Let these two values be respectively denoted by W and Z , then will the general rule expressing the value of an estate eV AB S. r-i W+ Z , and the value of a given sum _x V-AB -W + Z. When , the lives are all equal , the value of the reversion , by substituting the values of W and Z , becomes V+ CC-C--CCC , orr xV + CC --C-CCC , according as it consists of an estate , or a given sulm .
But the solution of this , like that of the preceding problem , may be obtained without having recourse to any other .
In the first year , the value of the given sum will be S into a ' b-rn c-d .
a ' .
b-n d a'.-d .
m b-r c-d .
aa ' abcr abcr zabcr +a bcr a 'm d b-m .
a--az ' .
d. abcr + a-br in the 2d year it will beabcr= into -n , e aI d-e .
n. , a " mn d-e .
a-ata M-n d e.a 1m-n.e.a ++2 --2~~~~~ +a no +4-n i e.a-a'+ a"c -d rmn a " C-d .
mn .
a ' b-rnm c-d .
a ' S7+ s- ; in the 3d ye are into ii ~-~o .
7'271.r"'+n-'o~f a " ' e n-f .
o. e-fia-a+ '+a ' n_o e. a nr 0-f z+2++ a ' .
of + n.-o.f.'a a ' + a"+ a"'+ c e. n-o.a"+a + a"c-e . ?
l- } bn.ce .
" ' if~ +"-e - ?
z2L , and so on in the other years .
If these several fractions be expanded , they will form nineteen different series , whose sum may be found =S into r 'x VABC- .
A+B _ AC+BC L AB+ x AK+ BK-2ABK 2z r2 R. AF-AFC M d.i+APT 2zb ^ 7r'.C i+ AP .
If B be the oldest of the three lives , let vr denote the probability that B dies after C , * then will the value of S , after the extinction of the life of B , be sqx_ and the whole a rx r value of the reversion will be=S into xV ABC A'+ B A ' C'+ BC .
AF_ AFC _F +AB _ , _ A " K+ BK 2z 2r2 b2 _ - , +APT BI , 2ABK + +CA -i+AP+ ?
aZr xr-1.V 'Aq , A ' , A'/ C ' , &amp;c. denoting the same quantities as in the former part of the preceding problem .
If C be the oldest of the three lives , let the symbols be changed as in the preceding problem , and the value of the given sum will be =S into xV -.ABC 2 A'+ B'b _______ m a. H'B'-HBC mS +NP AC+ BC m 2a Ijabr + NPC A zrB+AB B zbr X1+ PC ; +5rx+ NC ... *After the decease of C , the ' By the Table , page 229 , Phil. Trans. for the year 1794 .
given sum may be received , provided either of three events shall happen ; ist , if A shall have died after C in the first z years ) and B dies in the z+ 1 .
z+ 2 , &amp;c. year ; 2dly , if B having died after C in the first z years , A dies in the z+1e2+ 2.1 &amp;c. year ; 3dly , if both A and B having survived the first z years , the survivor of them dies in any of the following years .
Let vr denote the probability that A dies after C , and p the probability that B dies after C , ( both found by the Table in Phil. Trans. Vol. LXXXIV .
page 229 ) and let all the other symbols be the same as in the latter part of the preceding problem , then will the value of S , after z years , be =S into -q rix VA +a rx+1 +p r-t x ' V-B nr , + , x Bz +Az +Bz-aqx A. , + Bz -- .
ABz , and the whole value of the reversion will be Sinto r-I XVABC_I AF A. C H.E'-H FCr Zb 2a rn I+PC s. i+NPC AC+BC + AB +sl+NC Ibr &gt; '~~ za 2zr 2ar grq.r- , I A Pp.r-i rI arXrrb r+X ' ' ab r-+ V Az + BxABZ.* ab r-+ If the three lives be equal , the first of these rules becomes S intorr x V-C + CCCCC +-$Tc x CKCCK + d~d ~d dd-r x1+ CTT .
- .
xcr xI+ CT ; and the second =S into rx V-CC+CC .
CC cx CK-CCKccr xi c+ CTT c.x 1+ CT ; but the three last fractions in each of those rules destroy one another , and therefore , in both of them the value of the reversion is rX c+C -CCC .
&lt; .D Q.E.D. PROBLEM IV .
To determine the value of a given sum , payable on the death of A , should his life be thefirst or second that fails , and should B 's life , if it fail , become extinct before the life of C. Solution .
In this case , the payment of the given sum can only be prevented by the contingency of C 's dying before A , and therefore its value is immediately found by the solution of the 2d problem in my first Paper on this subject , * being no more than " , the Value of a given Sum on the Death of A , should C sir " vive him .
" The accuracy of this solution will appear from the following investigation .
In the ist year , the given sum will become payable should either of four events take place , the probabilities of which are expressed by the four fractions a'.b-m.c-d + d. b-m d zabc + abc d+ 'd LLL + a ' In the 2d year , it will beazb cza -b czac 2zac come payable provided either of six events should take place , ad m-n .
d-e as .
m-n.e an expressed by the fractions a+ab cT abc a. die.n b-m d-e , a b-m..ae a. Od a " e +.__aU _________ -+ z za cX zabc T abc 2ac a 2ac the third year , the payment of the given sum will depend on the same number of events , and the probabilities of the several contingencies may be reduced to ae a+ fr The value , thereSzac a/ '.d " fore , of the reversion will be X r+ r ' +3 &amp; C. S a ' die +SX a ?
+ a-e + ?
&amp;c. which two series are known zacrt to express the value of S , on the contingency of C 's surviving A.* WeTee a further proof necessary , it might be observed , that the value of the given sum , in this problem , is equal to the sum of the values of the two reversions depending on the con- , tingency of A 's being thefirst that shall fail , of the three lives ; and on the contingency of C 's surviving A , in case B shall be then dead .
Supposing the three lives to be equal , these values will be .
r-X - , CC..CCC , S X. rI Wr tr X_ S. ; rI-i xV CC , or the value of the reversion depending zr on one life 's surviving the other .
Q. E. D. PROBLEM V. To determine the value of a given sum , payable on the death of A , should his life be the second or third that fails , and should B 's life when it fails , become extinct before the life of C. Solution .
The value of the given sum , in the ist year , will be d. b-m .
c-d a b-med into 2+ 2 ; in the 92d year , it will be S a. m-n .
d-e a m-n.e abcr2 into 3+b m. e. a+ b~-m c-d-d a ' h b-m .
e a " + -m-d in the gd year abcr3 into 3n n~o .
+ +b-n.a " ' eef+b-n a"'.f + bfl .
cz-e .
a ' , and so on in the other years .
These several fractions being expanded , will form seven series , whose sum will be S into r.VA .
BC-ABC x BK br 6r 3c~ 1K ____ r. PC-IAPC d ABK + b. +_ __ _6-d-X BT -ABTm .
PT-APr 3 FC-AFC BC+AC-AB + -X 3 F-AF +-r + X. K-AK 2C When B is the oldest of the three lives , let x be the number of years between the age of B and of the oldest person in the table ; p the number of persons living opposite the age of C ; and q the same number opposite the age of A , at the end of x years ; A ' , A"C , and A"K , the values of annuities on those single and joint lives for x years .
The given sum may be received after the first x years , on the death of A , provided C shall have died after B in that time , or if C shall have survived the said term of x years .
Let r denote the first probability , * and tc will denote the second ; then will the value of this part of the reP q. r.V-Ax , n h hl version be S. into +xa rx+l andthewhole value of the reversion will be = S. into lX 2VA'-ABC I Found by the Table in Phil. Trans. for the year 1794 , page 229 .
2ZBK.~~ABKW 6X.AFK in d.i+P -3CX B -ABK +6b c+ 3br xPC-APC I+ T le ~~~~~~~d +X FC-AFC+ 3AF x BT ABT _Z-C X A'K ' A/ C'_AB P q.rI .
VAx +2r+7+cX arx+ When C is the oldest of the three lives , let q denote the probability that C dies after B let z be the difference between the age of C and that of the oldest person in the table ; A ' , A ' F ' , and A ' B ' the values of annuities on those single and joint lives for z years ; and k the number of persons living opposite a life z years older than A then will the value of the reversion in this case be= S into r6-2 XV LV '-ABC _X BK 3C AB + xAF me CP d. +APT +6 bc~ XAK+ 3br2 PC-P !
J _d 6cr X BT ABT + 6b XFC -AFC + 3AF2X ACA'B ' qk r_1.V Az AK + -+ ; r fA be the oldest of the three lives , let b ' , bl , b " ' , &amp;c. be substituted for b-rn , r-n , n o , &amp;c. and as , s-t , &amp; 6 for a ' , a " , &amp;c. then will the value of the given sum for the 1st S a-s c-d .
b ' a-s -db'i year be= x -cd +--i for the 2d year= ab cr 3X 2S s-t.de .
b " s-t .
eb " abcrl X+_ -+s--t.d e.b'+ s -t.eb ' 3~ ~~~~~~~~~~~3 +rcl and so on for the other years .
Hence the whole V le may be found =S into r-il x 9 , V 3A -ABC_ x* By the Table in Phil. Trans , for the year 1794 , page 2z9 .
F !
a ~ z~HBK - &amp; d xAK ABK ax HBK + 06 XHC-HBC dX AT S. I-4-NBT S NC-NBC AC_ AB ABT+s a--1a3 NB+ AWhen the lives are equal , the value , by the two first rules , will S.._ __ be = 6r X 2V-3 C-CCCT+lSfintoX .
K. -CCK xx me ~ ~d -a dd A tCrdK + 6cr CT CCT X1+ CTT ; and by the third rule it will be =S into 0LF x 2V 3C.CCC 6cc X CKK + 6rX CT , + crX CCT +2 6cc F6cr xF3cr m cr dd 6c crx 1+ CTT .
In both cases , all the fractions after the first destroy each other , so that the general rule is =S r2V ACCC , which may be proved , from other principles to be the true value .
As the reversion , in this problem , consists of two parts ; 1st , of the contingency of receiving the given sum on the death of A , provided B should be then dead and C living ; and 2dly , of the contingency of receiving it on the death of A , provided B should be the first , C the second , and A the third , that fails , it follows , that the present solution may be attained from those of the problem in my second Paper , * and of the 6th problemf in my last Paper on this subject .
But the computations derived from the addition of those two problems would be too tedious and complicated , and therefore the preceding rules are preferable .
In the particular case of the equality of the lives , the general rule becomes the same as above , ( or =S x 2V -3C CCC ) and consequently affords an additional proof of the truth of the investigation of the three problems .
PROBLEM VI .
To determine the value of a given sum , payable on the death of A , should his life be the first or last that shall fail , of the three lives ; and should B 's life , if it fail , become extinct before the life of C. Solution .
The value of the given sum , in the ist year , is absr into a.b-m.c-d ?
'd a ' .
b-m d a ' c-d .
m 2a m~d + in---+ --2 , nthe 2d year , __ ~~~~._ S. a".rn-n.d-e , , a.rn-ne1 its value is abcr into +a no +mea d-e .
n b-mr .
d-e .
a ' b-m c-da ' in the 3d year , aS 2_ +a+ ab cr3 into n-o.e-f aof + a. n-o.f a. e-fo + bn of .
a " ' 2+ ~~2 +2+ b-n .
c-e .
a ' , and so on in the other years .
The sum of these several fractions , after reducing them into their proper series , tor-i AC%AB may be found =S into xV -A-ABC + Az +X1+ APT-2 X1+ AT + BT1'-ABT-X zb cr2 cr 2hC AFK +x AF + FC AFC .
When B is the oldest of the three lives , let x be the difference between his age and that of the oldest person in the table ; and , as the given sum may be received after the necessary extinction of his life , either on the event of C 's having died after him in x years , and A 's dying in the x+ 1 , x+ 2 , &amp;c. years , or on the event of C 's having lived x years , and A 's dying in the x+ 1 , x+ 2 , &amp;c. years , it is obvious that the preceding rule in this case will not express the whole value of the reversion .
Retaining the same symbols as in the first case of Prob. V. the value depending on the former of those two events will be r. ql V-AX _i V-i , and on the latter of them it will be a real .
AC d __ _ ______ ACd ATx+ pq.r I.V-_Ax pq.r-{.VACx A PI -L 2l r zcro+l acr+Z 2acrx+I 9 therefore , the whole value of the reversion will be =S into r-I ~~~ ~~~AC AB mdrX V-A'-ABC +X -I2 + APTr Xr x2 r bcr Acr x1+ AT +BT-ABTc xAFK + xAF +FC-AFC pq.rx .
VACx Lp q.rXI .
V-Ax + FC AF 2acrx+I +x rx+1 When C is the oldest of the three lives , let p denote the probability that C dies after BX and let z be the difference between the age of C and that of the oldest person in the table ; A ' , A'B ' , and A'F ' , the values of annuities on those single and joint lives for z years ; and q the number of persons living at the age of A after z years ; then will the value in this case be =S into r-x Ad AC AB ' mdd zrx VA'-ABC+ + -2br x1+ APT-A rx1+ AT + BT-ABTh -3X AFK + __ X:zb cX ''-bF + q.Xr-i.V-At a r%+1 _ When A is the oldest of the three lives , let the symbols be changed in like manner as in the corresponding case in the * Phil. Trans. for the year 1794 , page zz9 .
preceding problem ; then will the value for the ist year be mS t sc-d.b a-sb-b.d+ a-sdb ' abcr 2a -z vs c~~d , b~~b ' .
s t. d-e b+ a-s .
c- ; d.b-b ' for the 2d year= abcinto 2+ s-t .
s-t.d-e b ?
bU+ s t.d e.b ' 2 ~~~~~~~~~~2 + s-t * c-d .
, and so on for the 3d , 4th , and remaining years .
These several fractions may be expanded into twelve different series , whose sum may be found S into xZ XV A-A BC 2zr AC AB AK a.HBC d3F2 zr 2C2az cr 2ar ~~ds X1+ NC + NBNBC r. + NBT .
Ithe lives be equal , the value , by the first two rules , will be t. VC+CCCCC dd lL .
Jd Sinto r v+ cX1+ CL cX 7zr , c cr 2c rx CKK 1 +2CI7-CCT-~tC 4-cx2CK CCK ; and by the third rule it will be S into V-C+C CC(+J zr 2C -~d -dd CCK-CK + xi + CT x1+ CTT .
In both cases , all the fractions after the first destroy each other ; so that the general rule becomes =rxV -C + CC-CCCC , which may be proved to be the true value , from other principles .
The solution of this problem might have been derived from that of the first problem* in my third Paper , and from that of the 6th problemt in my last Paper on this subject , " ; by finding the value on the death of A , should his life be the first that " failed ; and also the value of the same sum , should B be the first , C the second , and A the third , that failed : " but the foregoing rules , in being more simple , are preferable .
In the particular case of the equality of the three lives , the value , by these problems , will be =S into L2 W CCC +3r 6 , rXV-S3CWCCCCC= zr Xr V-(+CC-CC as before .
Q. E. D. PROBLEM VII .
To determine the value of a given sum , payable on the death of A , B , and C , provided C shall die after one life in particular , ( A ) .
Solution .
In the 1st year , the value of the given sum will be SS -2 X b-m .
c d. da ' ; in the 2d year , it will be= bc r i-to Mn .
d -a " --c-d mn a ' l~t~ a2+b-mde~a/ +A-+m-n b-m .
d-e .
a " ' S* n-o.e f. a " ' d -2e .
a ; in the 3d year =a into c. ? .
at + b-n .
e-f .
a+a+ 2-+ na-o .
e-f .
a'+ a " 2 b-n .
e--f .
a " ' +2 -- ) and so on in the other years ; whence the whole value may at last be found S into x BC -B-BC BC AB Jmd i+APT _ ABC ?
K2 2r + .
+At BK -ABK + R , ( R denoting the value of 5 , by the 3d problem in iny first Paper , on the contingency of C 's dying after A).This general rule gives the true .
value of the reversion , when B is the oldest of the three lives .
But , .
when C is the oldest of the three Ies , the general rule will be = SJ into rX BC , W BC A'B ' _-_I -ABC +z+ ( XI-APT gR_ AKBK + R+ vP71b rX -B x denoting the difference between the ages of CGand of the oldest person in the table ; p the number of persons living at the age of B after x years ; B ' , A ' B ' , and A'P ' , the values of annuities on those single and'joit lives for x years ; andk the probability that C dies after A.t When .
A is the oldest of the three lives , let the symbols be changed as in the solution of the preceding , problem , and the Sbb whole value of the given sum will be = zX b+ r3~~~~ ~~~ r cJ ~ rr +rC+bX +.bill 5 cb ' d b " bill 5 d~bf ' r+b r+ + r3 2 bcr X eb'+f.b S d_ bS b " f b+ ' s rz + &amp; C* bec X+ +7T bcr7-ei f.b+b"l S arb .
' s b ' ; tb " ' Ss cb ' dt b " eu b " ' Sdt b ' eu b. b " 2.abc X r-+ z+~ r3 &amp; C. 6 -2a x r+ be .
+ xC .
&amp; C.be ?
, &amp; )c .
+ 25b X r+ + ?
t r3J c+2absrX+ tf.b+by S sdb ' dteb 1f b"'C S + ; &amp; ) C. + 2 , acxr + r+ r3C+ +2 te-b ' uf .
b-+bzbc dsb ' t e. s " &amp; C_ , +._ : G. 2r &amp; rl + eC .
After the extinction of A 's life , the given sum may be received on either of three events ; ist .
on the death of B in the z +1 , z+ 2 , &amp;c. years , ( z denoting the difference between the ages of A and of the oldest person in the table , ) C having died after A in the first z years ; 2dly , on the death of C in the z+ 1 , z+ 2 , &amp;c. years , B having died in the first z years ; 3dly , on the extinction of both the lives of B and C after the first z years .
Let p denote the probability that C dies after A in z years ; * p the number of persons living opposite the age of B ; and k the same number opposite the age of C at the end of z years ; and the value on the two first of these contingencies will S bpp .
r.V-B--+ S.b-p.k.r-i.V-C.x be -.i f+b r~ Agains let b r~7+1 +b c rz1+1.Aan e greek letters be substituted for the corresponding italic letters in tlthe first part , and the value on the third of those contingencies S. S Will be bf r+ rz + 3 , &amp; 5C .
-bc r--X r+ rzn &amp; aI@+ + , S. g+ s~j3~ " C. &amp; C. br+ .
X +R+ ( j'jl , + , \ Jon &amp;c. The 1st of these series being added to the first series-in the former part of the solution , their sum will be=S into r--oV-B +-zb~c , .
; ' the 2d and 4th being 2zr + brx+1 ' added to the 4th and 5th , their sum will be =S xC BC ; and the gd series being added to the 3d series inthat part of the solution , their sum will be = sSozr C ; so that these six series r - .
r_ _ BC last mentioned are _r_ The whole value of the reversion in this case may therefore be fouind = S. into V V-B C LB+ LBC + AC ABC + IG ' z r:~~~~~~z r+ za HB + H-C BC -C XBK +AK-ABK + D+ X p__ ____ b-P l-r k~ %.BC -I rV-B b+ rx , ,an4J K ' , denoting the values of annuities on those joint lives for z years .
When the three lives are of equal age , the value by the first two rules will be =S into XV -FC.tCC'-CCC +X+ CT x+ CTT x-xCK .
CCKin 2zcr 2zc crX +iwhich the last three fractions destroy each , other ; and by the last rule the value will be-= *'x VAC+3CC-CCC ; that is , , in both cases , " half the reversion after the extinction of " the three lives , " which from self-evident principles is knownto be the true value .
The solution of this problem may also be derived from those of the 3d problem in my first Paper , * and of the ist problem in my last Paper , + c " by deducting the value of an estate after the " death of C , provided that should happen after the death of A , " from the value of an annuity on the life of B after C , pro"vided C should die before A."Thus , in the case of equal lives , the value by the first of these problems being v-_ C+cci and the value by the second being = 2_ , their difference , or V-CG is the number of years pureldest of the three lives , has been investigated much in the same manner with the prep .
-sent case ; but the operation was omitted merely for the sake of conciseness .
chase required , and consequently the value of a given sum is Sx 2r xV3C+3CC CCC , asbefore .
But the rules derived from the foregoing solution are in general more simple than those derived from the two problems just mentioned , and are therefore to be preferred to them .
The foregoing problems , together with those which have been investigated in my former papers , comprehend , as far as I can perceive , all the different cases of survivorship between three lives .
The great number of contingencies on which these reversions depend , must necessarily render the solutions intricate , and consequently the general rules complicated and laborious .
It would not , however , be a difficult task to abridge these rules very considerably , without destroying their accuracy in any great degree ; but this would be foreign to my purpose in these papers , which has uniformly been confined to the investigation of the correct values of the different reversions .
Nor do I think that such an abridgement is necessary , as the opera'tions of even the longest of the present rules , may be completed.in very nearly as short a time as the inaccurate approximations which have hitherto been employed for the same purpose .
It may not be improper to observe , that the solutions in these papers are not only the first which have ever been deduced , in the case of two and three lives from just principles and the real probabilities of life ; but that , as to many of the problems , not even an attempt has ever been made to approximate to the value of the reversion .
Being now possessed of correct solutions of all the cases in which two and three lives are involved in the survivorship , we , are possessed of all that is really useful , and therefore I feel the greater satisfaction in closing my inquiries on this subject .
For , in regard to contingencies depending on four or more lives , the cases are not only much too numerous and intricate to admit of a solution , but they occur so seldom in practice , as to render the entire investigation of them , were it even possible , a matter of little or no importance .
Ln my last Paper , printed in the Phil. Trans. for the year 1794 , page Z27 , last line but one , fr B WC-B red Pkr-i , V-BCz % BC ' read P rt .
-sZ ; whence the general rule i the intoLI xV-3~.AB B+ C AC j folluowing ' page becomes _ Sinto xV-3 BC + CA_ X AF-AFC HFC ca x.HB-HBC s 3-+ a+ 2a X I-IC +x NB .
, *NBC m. e +NPC m z.AP-APC p_ _-VC b xbr 3 ++b Xr pk .
r- .
V-BCz zbcrx+t

