]\gt ; An of the of Probabilities to the Study of a priori thometry .
\mdash ; Part II .
By Lieut.-Colonel Sir RONALD Ross , K.C.B. , , and HILDA P. HUDSON , M.A. , Sc. D. ( Received October 19 , 1916 .
) CONTENTS .
PAGE .
Errata in Part I 212 Preface to Part II 213 SECTION VIII .
Hypothetical Epidemics .
Deduction of Constants 213 ( ii ) Examples with Table 215 ( iii ) Detailed Calculation of and 219 SECTION IX .
Hypometric Happening : KL negative 220 ( u ) The Happening ceases : 222 ( iii ) lies between zero and 222 ( iv ) lies between and 223 ( v ) Isometric Happening : 223 ( vi ) Remarks 223 SECTION X. ( i ) Parameter anadysis of and 224 EBRATA IN PART I. Proceedings , 211 , line 23 , ?
for 213 , equation 17 , omit factor .
equation 18 , factor .
line 23 , ' : since is greater than I 215 , line ) , for line 21 , ?
urite 217 , equation 29 , write .
218 , line 22 , for line 2 from bottom , ?
orite 219 , equation 36 , for equation 38 , ( twice ) .
equation 39 , write .
for and for Study of a priori Pathometry .
Page .
221 , line 8 , last term , ?
for L. lines 4 , 3 from bottom , write " " seen that all the tangentials of are zero 222 , lines 6 , from bottom " " nearly \ldquo ; , line 10 , for 225 , equation 67 , writc ( 1-L ) .
229 , equation 80 , for 230 , 3 , Part for Part To PART II .
Part I of this paper was published in the ' Proceedings , ' , vol. 92 ( 1916 ) , having been read on November 11 , .
In June last , the loyal Society was kind ] to give a Government Grant for providing me with assistance in order to complete the papel , and for carrying on further studies upon the subject ; and Miss Hilda P. Hudson , 1f.A .
, Sc. D. , was appointed for work from May 1 , .
The continuation of the paper has accordingly been written in conjunction with her ; and I should like to take the opportunity to express my obligations to her for her valuable assistance , especially in regard to Part III\mdash ; which is to appear shortly .
I must for the rather numerous small errors in Part I\mdash ; due to the fact that the proofs were received by me when I was abroad on active service .
The entire paper here presented is still limited to the theoretical side of the subject , as defined in the Preface to Part I. Records of epidemics are now being examined in order to find how far the theoretical results which we have reached may be applied to them ; but these studies must be reserved entirely for future discussion . .
R. VIII .
( i ) Hypothetical : Deduction of \mdash ; In any case of independent hap we can use equations 8 and 12 with , all zero .
Then , and , , if is small , for all values of 214 Lieut.-Colonel Sir R. Boss and Dr. Hilda P. Hudson .
If the value of is known for any value of , say a year , this eq.uation enables us to find , and thence for any obher period , hour , day , week , etc. Thus suppose that child-birth , death , or ration occurs to 20 per of a population of all ages during one year , to what proportion do they occur in one day ?
We have To find the proportion for one hour , we divide by 24 ; and to find it for 7 or 30 days , we multiply by these numbers respectively .
Or suppose that 85 per cent. of plague cases , taken at any stage of the illness , die within 30 days , and we require the daily death-rate of such mixed cases .
Then we have , giving a constant death-rate of per cent. per .
If the cases are all taken from the commencement of the illness , the curves are quite different , and show the highest death-rate some days after the onset\mdash ; but this does not concern us at present .
If the annual birth-rate is 30 per mill , and the annual death-rate is 20 per mill , then the daily are and Hence if the ration and emigration rates counterbalance each other we may take for the daily variation element In this example the increase of the population at the end of the year would amount to 1 per cent.\mdash ; that is , would vary from at the beginning of the year to at the end of it .
But if we use the value of just given for equation , namely , putting and , we obtain at the end of the year\mdash ; which is a little greater than the figure just given .
The reason for this is , of course , that in equation 13 we assume that the propagation is continuous\mdash ; that is , that during the year the of the original population will themselves have .
But this is not assumed in the value of first obtained .
We may also calculate from equation 13 if we know the values of and at the beginning and end of .
Thus the population of England and Wales was in 1811 , and 36,070,492 in 1911 .
Then since and , the daily variation-element is , the yearly variation-element is , and the ten-yearly variation-element .
If a population doubles itself in a period , we Study of a priori 210- have , so that at the above rate the English population ought to double itself in 19,975 days , or nearly 55 which is about the case .
If the daily death-rate of is taken at the figure just calculated , of , and the plague birth-rate is a normal one of 30 millc a ( so that ) , then ; so that , if daily .
Such ures give us some concrete ideas of what the constants are likely to be in cases of hunlan epidemics .
[ Compare Sections III and ( iii ) .
] The reversion-element .
is particularly elusive since we seldom have known figures it to deal with .
It is best to proceed as we did in connection with equation 20 , and to assume that 90 per cent. of the affected individuals revert in the time , where , 2 , 3 , , 10 , or more years .
Then for a daily element .
Hence if 90 per cent. of the cases revert in one year , the daily reversion-element is .
Thus the reversion-element is likely to be compared with or and corre- spondingly important , and will be by no meanJs nible even when the considered disease is one in which the acquired immunity is supposed to be very lasting [ compare Section III ( iii ) ] .
For a case reverts not only when it becomes tYain capable of showing a recognisable infection , but when it is again able to harbour the infective ents sufficiently to them a whence they may infect others ; and it is quite possible and indeed probable that this may often occur much earlier than we .
Thus persons vaccinated ainst smallpox may acquire the disease in a very mild and modified form after only a few years , and may then infect non-protected persons in full force ; and there is no proof that a mild first attack of measles or other diseases may not give quite as short-lived an immunity\mdash ; that is , that those who have had the disease in infancy may not compara- tively early acquire it again in an indistinguishable form and then spread it , .
In such cases , may be considerably larger than we migbt otherwise expect .
Regarding immigration and emigration , it will suffice to note that in 1911 there were 350,429 immigrants into the United Kingdom and 454,527 emigrants from it .
Further remarks on the constants will be found in Section X. ( ii ) We may now proceed to consider numerical examples of proportional happenings .
For this purpose it is convenient to adopt the suggestion made towards the end of Section , and to take as a function of an independent parameter which is such that when is positive KL is positive and always increases with the time , and when is negative ICL is 216 Lieut.-Colonel Sir R. Ross and Dr. Hilda P. Hudson .
negative and always diminishes with the time ( Section IX ) .
We shall also write for and for .
The principal constants of and may then be written The reader is reminded that KL is the coefficient of in , the proportion of affected individuals ( equation 50 ) , and that the -curve reaches its centre of symmetry when , and its maximum value when is very large [ Section ( iii ) ] .
The period at which reaches its centre of symmetry and reaches its maximum is .
If and is large compared with unity , then , as suggested in connection with equation 58 , we may now write , and at twice this period the maximum number of affected individuals less one is reached .
Similarly , when is very , the ultimate value of the proportion of new cases is .
But if rises to a previous maximum value when , and thereafter falls to its ultimate value [ Section ] .
The ratio between the maximum and the ultimate value is ( equation 67 ) .
The peliod at which the maximum value is reached is ( equation 66 ) ; and when and is large , this may be written .
But this maximum of ( previous to its ultimate yalue ) occurs only if It is evident that when is expressed as a function of in this manner , , and are independent of and , and that contains them only as parameters of On the other hand , and are functions of the original population and increase as it increases\mdash ; because , obviousJy , and require a longer period to reach their values if the population is large .
It is convenient therefore to adopt a definite figure for in the following Table ; and we put , which is , say , the population of an ordinary town .
We may also assume that unity is negligible compared with unless If is not 10,000 , but some multiple or fraction of it , say 10,000 , all Study of a prPriO Pathometry .
218 Lieut.-Colonel Sir R. Ross and Dr. Hilda P. Hudson .
we have to do to find and is to add to or subtract from the figures given in the corresponding columns of the Table the quantity Thus , if , we subtract .
As increases above 100 , and approximate more and more closely to the value being always less than this and freater than it .
The actual curves depend largely on the values of R. If we suppose the case of a human disease , of which the affected birth-rate equals the natural birth-rate , which is taken at daily , corresponding , according to the previous sub-section , with an annual birth-rate of 30 per then equals the former if is so small as to be igible .
But if is a daily rate corresponding to an annual reversion of 90 per cent. , it will amount to as much as , so that .
We will suppose that these two values of are a minimum and a maximum .
Now if we suppose also that , that is , that each affected individual daily infects or reinfects one other individual then we have That is , if if is a minimum , and or } if is a maximum .
Thus , if the infection rate is near unity , the value of will reach from three to five figures , and the epidemic will resemble the last examples yiven in the Table .
In Example 1 , if the population remains the same during the whole of this period ( which is , of course , never likely to occur ) , only about .
ten living persons out of the 10,000 will be found unaffected at the end of and the ultimate and maximum daily number of new cases reaches only , that is , only one new case in years .
If , however , has its maximum value , there is one new case in about sixteen days .
In Example 4 , half the population is ultimately afl.ected , but still has no maximum before its ultimate value .
When , however , in Example 5 increases above unity , this maximum to appear , and the ratio between the maximum and ultimate values of constantly increases with At the same time , diminish and approach each other .
If is nearly unity , days , and reaches its in a short time , whatever the value of B. More generally , if , and if is large ( say over 50 ) , , and .
In othel words , if is considerable , not only does nearly , but the maximum of and the mode , or time when that .
maximum is attained , depend roughly on only , the population being fixed .
If the -curve slopes downward from its summit more slowly than it towards that summit .
Put , then , and has the same value for equal and opposite values of Study of a priori Pathometry .
But ; , and is less when is positive than when has the equal and opposite value .
Hence , at two points of the -curve whose ordinates are equal , the side is less steep than the ascending side .
Since , and the first tcrm on the gives a perfectly symmetrical bell-shaped curve , we may call .
the circcss of the -curve .
When and is considerable , the excess varies from at the centre of symmetry to , when is very large\mdash ; that is , roughly , from to .
Both of these become small when \ldquo ; the reversionrate , is small .
The important result , therefore , follows that , if also is near umity , and is large\mdash ; that is , if the infection-rate is while the reversionrate is low\mdash ; then the -curve becomes a nearly nletrical bell-shaped curve .
This , according to Dr. Brownlee ( Section I ) , is just the kind of curve presented by epidemics of certain zymotic diseases .
The inference is therefore that such epidemics may be wholly or chiefly mere cases of proportional happening , as defined in Section .
It is even the case that , in such diseases , the fall of the curve is often nor slow than its rise , as happens with the f-curve .
When is and nearly , the two changes of curvaCure of the curve occur , to equation 65 , when , that is , when equation 53 ) , and the latter being nearer to the summit than the former .
( iii ) Several methods may be employed for the curves of and in detail .
The simplest is to divide the period during which varies from to into a number ( say 10 ) of equal parts and then to calculate the ordinate of the curve at each section .
In equation we used to express this interval when , bub it may be employed more generally to express the abscissa of the ceutre of symmetry , when has any value , so that .
We now put , give to the successive values 1 , 2 , 3 , ..- , 10 , and calculate the corresponding values of , namely , , and so on .
Thus we have from equations 50 The root can be easily evaluated by means of logarithms , and the value of Lieut.-Colonel Sir R. Ross and Dr. Hilda P. Hudson .
then obtained from a table of reciprocals ( e.g. , as in Barlow 's Tables ) .
Owing to the symmetry of the -curve , its values when lies between 10 and 20 can be easily found\mdash ; for example .
When we have ascertained the ordinates of , we can quickly calculate those of from equation 61 , and , unless the case-mortality is high , we can generally assume for short and sharp epidemics that the population remains constant and that and .
It will be observed that , by this method , when has been first calculated we need subsequently deal only with the values of and of .
Thus suppose that nearly , that and that , then we have approximately , This is , of course , an ultimate case where is quite symmetrical\mdash ; so that , for example Another useful method for and is to the origin to the centre of symmetry ( equation 56 ) .
But if the ordinates are required for uccessive natural units of time , such as days or weeks , in order to compare them with statistics of epidemics in which such units have been , then we must of course yive to the successive values 1 , 2 , 3 , , and calculate ordinates directly from equations 50 .
Here the symmetry of will not be so useful , because the centre of symmetry is not likely to coincide with any one of the periods of time for which the ordinates have been found .
It will , therefore , be generally easiel to plot the curve by the method first indicated , to measure off the natural units of time along the time-axis , and then to estimate the corresponding ordinates geometrically .
IX .
( i ) pometric H : KL iVegative .
\mdash ; We have hitherto always supposed that the happening-element is constant during the whole of the period , but it is now necessary to examine the changes which will be produced iu the fimctions by certain changes in this element .
We here study only the cases of dependent happenings already discussed , and these only for hypometric .
It is no very useful to take as a function of In Section , while considering the constants , we saw from Study of a priori Pathometry .
equations 50 that if is positive , always increases and is therefore always greater than , its value when .
This implies that .
When this inequality becomes an equality , KL vanishes and , that is , remains near its original value , when is not very large , so that the may be said to be ?
omefric [ see ( v ) below ] .
When , however , becomes ative ; and it is seen from equations 50 that , as then always increases with the time , always diminishes , and must therefore be always less than We call such cases .
That is , we now suppose that , after having been greater than becomes less than it , or even zero .
What , then , will be the nature of the curyes ?
The solutions of equations 44 , 45 , 47 , 51 , 52 remain valid for all real values of , and we have therefore only to interpret results already known .
It is first necessary to examine , and more larly as functions of but we shall do so only in the case when is positive { injurious ) or zero .
Then are positive constants .
We have , KL represented by parallel straight lines meeting the axis of at an of on the positive side of the origin , , or and the graph of is a rectanguJar hyperbola , with asymptotes , dotted lines in the figure , parallel to the axes of co-ordinates .
The last portion of this curve , for , gives cases of hypermetric happening already studied ; but we } ) to consider the range : since becomes nite when , we have to consider the two portions : 222 Lieut.-Colonel Sir R. Ross and Dr. Hilda P. Hudson .
, for which is positive and greater than 1 ; and for which is negative .
( ii ) First let ; that is , suppose that the happening suddenly ceases altogether , while all the other elements remain the same .
It is advisable to use marked letters for the quantities after this event .
We take it then that when the original happening ceases , , and that the values which , had reached up to this time are now expressed by , while denote these functions as increases .
Similarly , and are now the values of and without the element .
Tben , since equations 51 and 52 become , ( 81 ) ( 82 ) As decreases from its original value or and ultimately approaches zero\mdash ; unless the immigration is greater than the sum of the other elements ( as may happen in a war area ) its diminution is more rapid when are large .
With regard to equation 81 , we observe that is the natural increase of the population , lliving when the happening ceased , after that event ; and the equation shows that equals this quantity less a function of which is zero when , which increases with , and which finally slowly approaches the limit when is .
That is to say , the population which remained when the happening ceased continues to suffer from its effects for some time afterwards , and then finally reaches the value .
Here is always less than unity , for is the value of at the moment when the happening ceased , and this could not have been greater than which is less than unity [ Section ] , while , as shown in the previous sub-section , is greater than unity when .
Hence is always less than , and the difference gives the loss of population due to the cases which continue after the hap nceased .
The fraction occurs in the denominator of in equation 50 .
As it is reater than the coefficient of is positiye , so that always diminishes as increases\mdash ; as could have been already inferred from equations 81 and 82 .
( iii ) The next case is when is some quantity between zero and ( which is generally very small ) .
For the study of this case , the first of equations 50 may be written Study of a priori thometry .
which diminishes indefinitely as increases .
If and is consequently infinite , this becomes , which vanishes .
The values of and are easily obtained from equations 51 and 52 , if we remember that the indices and are now ative .
When , the expressions for and become indeterminate , but can be easily evaluated , and we have , where ( iv ) When lies between and is positive and is negabiye , and varies from to zero .
Then it will be seen from equation 83 that is still positive , and , as in the previous cases , diminishes indefinitely as increases .
For it may be written ( 84 ) where is negative .
For we have ( 85 ) which is also always positive since is negative , and increases without limit when is large .
is less than , because the effects of the original happening continue to be felt by the population after the happening itself has been reduced from to ( v ) When exactly , and , and , when is small , nearly , and the happening may be said to be isometric .
The values of and become indeterminate , but can be ascertained by the limits when of the expressions in equations 84 and 85 , so that Or the same equations can be obtained by integrating equation 47 after putting in it , and then integrating both sides of equation 44 .
In this case also , then , tends to decrease as increases , and finally approaches zero , when is very large .
( vi ) If is positive , however small it may be , always increases , as already seen in Section ( iii ) .
It must be remembered that , though , the proportion of affected individuals , diminishes , this does not mean that new cases cease to occur .
On the contrary , unless is absolute zero , is the same function of as before .
And expresses , as before , the reversions of the old cases .
When , the phase described in ( iii ) above is suppressed , and we 224 Lieut.-Colonel Sir R. Ross and Dr. Hilda P. Hudson .
, when , at the point when , and and the functions will bs easily understood .
X. ( i ) \mdash ; We have now considered , and as functions of the time , but it remains to examine how they vary if we give different values to the parameters ) , , 7 , E. The symbol has been used to denote the limit of when is very large .
This is an important quantity in disease happenings because it gives the number of new cases which continue to occur when has reached its limit , that is , after epidemic manifestations have ceased .
It is , by hypothesis , these final new cases which keep the infection alive permanently in the population ; we may call the endernic ratio .
First consider as a function of .
Thus is positive when and increases with until this .
eaches a value which makes and [ Section ] .
After this diminishes as !
increases .
But if then never reaches this value and always increases .
Thus considered as a function of has forms similar to those of considered as a function of The behaviour of as a function of is of the same nature .
A similar procedure applies to and P. The effect upon the total population due to variations in the parameters is important .
The fundamental equation is 44 : As and are always positive , the of depends on and D. If , the happening is equivariant and the natural change of population due to is not affected .
If is negative , the is beneficial and the natural change of population is augmented by it .
If is positive , the is injurious and the natural is reduced .
If is negative , especially when , the total population will diminish indefinitely .
It is found that is always positive ; that is , in infectious diseases , quicker loss of affectedness , .
loss of immunity , is beneficial .
seem until we rellect that , as , the longer the affectedness continues in the individual the greater will be the loss of life due to it , since we generally suppose in such cases and that these elements act , not one , but every lived by the affected part of the population .
Of course , the greater is the more quickly on the average will affectedness , including immunity , be lost .
Study of a priori Next , is zero or negative , and the higher the infection rate the more injurious is the happening .
is zero when .
It generally increases at first with and afterwards decreases ; it is negative when proyided negative .
In other words , in these cases an increase of is beneficial at the beginning of the outbreak , but injurious later on .
Now and diminishes if the case-mortality increases ; in fact Thus an increase in the case-mortality is injurious at first , but may be ultimately beneficial , because an affected individual who dies ceases to be infective .
Application of the Theory of : to the Study of a priori -Part III .
By Lieut.-Colonel Sir TALD P , K.C.B. , , and HILDA P. HUDSON , M.A. , Sc. D. ( Received October 19 , 1916 .
) CONTENTS .
PAGE SECTION XI .
( vi ) Assigned 232 ( vii ) More General Equations 234 SECTION XII .
( i ) Mortality .
First Modification of the Equ tions of Part I 236 ( ii ) Second Modification of the Equations 237 SECTION XIII .
( i ) Discussion 238 XI .
( i ) Variable Happening.\mdash ; In Section IX , we commenced by touching upon the necessity of studying the effect of changes in the happeningelement , but there dealt only with the case of hypometric happening .
We now proceed to examine other in this element .

