Square Roots by Hand

Instructions: 1 Count the number of digits your number has that are on the left side of the decimal point. If the number of digits is odd, then add a "0" to the front of your number. Ob iously that doesn!t change your number!s alue, but it!s important because no" you need to brea# the number into groups of t"o digits, and it!s important that you brea# your number up on the proper boundries. $ )oo# at the first group of t"o digits as its o"n number. *ind the largest number "hose s+uare is e+ual to or less than that. ,his is the first digit of the exact s+uare root. / 0+uare the first digit of the s+uare root and subtract the result from the first group of t"o digits. 2 3o" append the next group of t"o digits to the result of step / to obtain the partial dividend. - 3o" multiply "hat you ha e of the s+uare root so far by $, and discard any decimal points. ,his is the partial divisor. & 5i ide the partial di isor into the partial di idend and ta#e the first digit of the result. ,his is the next digit of the s+uare root. 0ince "e "or#ed "ith the first digit after the decimal in our original number 7the & in $%.&8 to figure out the ne" digit, "e put a decimal in front of the ne" digit. 0o far, our s+uare root is: 9 ,a#e the ne" digit from step & and append it to the partial di isor from step fi e. % :ultiply the result by the ne" digit. ( 0ubtract that from the old partial di idend. 10 ;ppend the next group of t"o digits 7008 to the result of step ( to get a ne" partial di idend. ;t this point the algorithm starts repeating steps: 11 <epeat step -: :ultiply "hat you ha e of the s+uare root by t"o and discard decimal points, to get a ne" partial di isor. 1$ <epeat step &: 5i ide the partial di isor into the partial di ided and ta#e the first digit of the result. ,ac# the ne" digit onto the end of the s+uare root, to get: 1/ *or more digits of accuracy, =ust repeat steps 9 through 10, and then loop bac# to step - as many times as you "ant. ;dd more groups of "00" as necessary. Example: $%.&, brea#s up into "$%" "&0" "00". 'o"e er, (.1 brea#s up as "0(" "10" "00".

,he first group is $%, "hich gi es us since -$ . $-, "hich is less than $%. $% 1 $- . / "/" 4 "&0" yields "/&0" - x $ . 10 /&0610 . /&, "hich yields "/"

-./ "10" 4 "/" yields 10/ 10/ x / . /0( /&0 1 /0( . -1 "-1" 4 "00" yields -100 -/ x $ . 10& -100610& . 2%.11..., "hich yields 2 -./2

As if all that weren't enough, there is one thing to watch out for. In fact, if you tried to follow that set of instructions with some other number than 28.6, you may well

have found out what that is: The result of step 9 may be a negative number! " o what!", you may as". #ell, since the results of ste$ % are the basis for the ne&t $artial dividend, getting a negative number means having a negative $artial dividend. And that, when you get to ste$ '2 (or the re$eat of ste$ 6), means that the ne&t digit of the s*uare root will be negative+ ",oodness+" you e&claim, "what on earth does it mean to have a negative digit in a number!" -ortunately, the answer to that *uestion is "nothing". If you thin" bac" to when you were in the first or second grade when your math teacher e&$lained the decimal notation system to you, you'll remember that decimal notation is a (very convenient) shorthand for a sum. -or e&am$le, '2..6 is ' in the '//'s $lace, $lus 2 in the '/'s $lace, $lus . in the one's $lace. 0r, '&'// 1 2&'/ 1 .&' 1 6&.'. 2he digits are sim$ly shorthand for various $roducts that you add together. o with that in mind, we see that having a negative digit is no big deal (aside from being aw"ward to write). It 3ust means you have to do a little math in order to come u$ with a "normal" number that has all $ositive digits. -or e&am$le, what if instead of '2..6, the 2 was negative! 2hat would be '&'// 1 (42)&'/ 1 .&' 1 6&.', which is '// 4 2/ 1 . 1 .6, or 8..6 o if the result of ste$ % is a negative number, don't worry about it. 5ust $roceed as normal. 6et's loo" at an e&am$le that has this $henomenon (I'll 3ust show the ste$s here, I won't re$eat the descri$tions of them): 2he s*uare root of ...'7 is: 0tep: <esult: s+rt799.1-8 . >.>>> 99.1- yields "99" "1-" "00" %$ . &2, and &2 ? 99 %.>>> 99 1 &2 . 1/ "1/" 4 "1-" . "1/1-" %x$ . 1& 1/1-61& yields % %.%>> 7. % 4 %x10118 "1&" 4 "%" . "1&%" 1&%x% . 1/22 1/1- 1 1/22 . 1$( "1$(" 4 "00" . "1$(00" %%x$ . 19& 1$(00619& yields 11 %.9( 7. % 4 %x1011 4 7118x101$8 "19&" 4 "11" . 19&1 ,he deri ation of that 19&1 is actually a little tric#y. <emember that you ha e to append the ne" digit to the partial di isor, then multiply the result by the ne" digit. ;ppending a negati e digit is a different operation than tac#ing a ne" digit on to the end of your result, because no" you!re really performing the calculation 71719&x10 4 188, "hich you then multiply by 11 to get 19&1. ,he minus signs nicely cancel out,

0 1 $ / 2 & 9 % ( 10 11 1$ 1/

12 11& 19

"hich is good. 0o "hat you really do is append the absolute value of the ne" digit to the partial di isor, and ma#e the result the same sign as the ne" digit. ,hen multiply by the ne" digit as normal. @ou "ill al"ays get a positi e result from this. 1$(00 1 19&1 . 12&&1 "12&&1" 4 "00" . "12&&100" %9(x$ . 19-% 12&&100619-% yields 1$ %.9%% 7. % 4 %x1011 4 7118x101$ 4 71$8x101/8

And indeed, that's not a bad answer since 8..882 8 ...228%99, which is only off by ./.8%99. :ot bad at all, for only three cycles through the algorithm. ;ou may wonder if this business with negative digits ever sto$s. 2he answer is no. 2he reason is because of the nature of the algorithm. 2his is a successive approximation algorithm, which means that each ste$ gets you a little closer to the correct answer. o if your initial a$$ro&imation is too small, the algorithm will have to add a small $ositive amount at each ste$ in order to ma"e $rogress towards the correct answer. <ut if the initial a$$ro&imation is too large, the algorithm has no choice but to ta"e small amounts away from the a$$ro&imation at each ste$. 0therwise, the a$$ro&imation would 3ust get worse. <ut how does it get li"e this! :otice that ste$s ' through 9, are different, and they're not re$eated. 2hey are engineered to always give you a $ositive first digit that is an under4a$$ro&imation of the true answer. 2his is not difficult to see= ste$ 2 tells you straight away to find a number whose s*uare is less than the first grou$ of digits. 2herefore, that number's s*uare root is less than the s*uare root of the first grou$ of digits. It is guaranteed to be the correct first digit because it is the s*uare root of the largest $erfect s*uare that's less than the first grou$ of digits, but nonetheless as a value in itself it is less than the true answer. ince it is easy to figure out that the first digit of the answer is an under4 a$$ro&imation of the true answer, it is also $retty easy to see that the second digit of the answer will be $ositive. 2he second digit has to be $ositive in order to bring the a$$ro&imation closer to the true answer. >owever, it is 3ust $art of the nature of the algorithm that the second digit can "overshoot" the true answer. #hen that ha$$ens, the a$$ro&imation changes from an under4a$$ro&imation to an over4a$$ro&imation, and so all your remaining digits will be negative. I leave it to cleverer minds than my own to figure out why it is $ossible for this second digit to overshoot the true answer (and ho$efully to fi& the algorithm). -inally, one last note. In the event that you choose to find the s*uare root of a number that has a rational s*uare root (such as '..69, whose s*uare root is e&actly 9.2), you will find that at some instance of ste$ % the subtraction comes out to e&actly ?ero and there are no more grou$s of digits left to deal with that aren't "//" themselves. If that ha$$ens, you "now you're done, and you have found the e&act s*uare root you're loo"ing for (not merely an a$$ro&imation of it).

2his is easy to $rove= if you assume that the result of ste$ % is e&actly /, and you follow that ?ero through the calculation, you'll see that all other digits $roduced by the algorithm will also be /.