Dynamics of Naval Craft - System Identtftcatton 
(footnote A). There are a dozen surface-piercing foil craft at Hong 
Kong, mainly made by the Japanese, but no country has a hydrofin 
for commerce (footnote B). 
I realize that my remarks are disjointed and I apologize to the 
very able Oceanics team. I am myself a statistician and I am enorm- 
ously impressed by the precision of the authors' predictions. C'est 
magnifique, but is it what we want ? I am reminded of a paper read to 
the Royal Statistical Society Journal twenty years ago - and this should 
should please Professor Telfer - by Barnet Wolf on multi-factor re- 
gression analysis. He called it the add-a-variant system for he added 
a factor at a time and tested for significance. Does the added factor 
improve your predicting of the criterion you are really interested in- 
namely, stability at speed ? If not, and if you carry on this testing of 
possible factors, you will end up with a sensible regression equation, 
having 5 or 6 inputs. Another method is to confine attention to a single 
factor with a sequence of observed values over time. We cannot afford 
to have clutter. We just simplify the problem and predict, whether by 
multi-factor regression or by auto-regression (footnote C). 
If I may have one more minute I should like to say that after I 
had used my firm, Plasticol, Ltd., to exhibit the hydrofin in London 
in January 1969 our Centre of Industrial Innovation at Strathclyde 
University decided to do a computer study as advocated by the Ocean- 
ics Team. The report ran to 66 pages and cost £10,000, but it did 
not produce the low-cost sea-transport. There was no man willing to 
risk his money in producing hydrofoils for the British Navy or for 
anyone else. So we turned in desperation to the Daily Express, whose 
A. The authors do not name their hydrofoil, but Mr Kaplan told me 
it was the (naval) Plainview of Grumman-Lockheed ; this craft has 
now joined the costly ladder-type hydrofoils of the Canadian Navy in 
retirement. 
5. My prepared comments had been handed to the Chairman, but 
they did refer to the re-designing of hydrofin by Cox & Gibbs in 1952, 
with sonic sensers, for the US Navy. Mr Hook has been asked to de- 
scribe the system at the January 1972 Symposium of the Aero-Space 
Corporation, Los Angeles. 
C. Auto-regression includes the Box Jenkins formula and my own, 
Orthodiction, with its freedom from swings with runs of errors in 
prediction sharing the same sign. 
1691 
