In 0rder　to　clarify　the　formation　mechanism　for　gas－phase　smoke　radicals　from　the　combustion　of　polymers，the inHuence　of　combustion　conditions　on　the　products　from　the　combustion　of　polypropylene　has　been　studied　using　gas chromatography（GC）．The　profiles　of　the　products　have　been　compared　with　those　of　gas－phase　smoke　radicals， As　a　result，a　correlation　betweell　oxygen－containing　products，such　as　an　a￠etone，and　gas－phase　smoke　radicals　was shown．Especially，if　the且ame　combustion　occurred，oxygen－containing　compounds　were　slightly　detected　and　gas・phase smoke　radicals　were　hardly　detected． Gas－phase　smoke　radicals　were　detected　from　the　combustion　of　acetone，an　oxygen－containing　compound．This suggests　that　meta－stable　compomds　such　as　a　trioxide，that　could　decompose　to　form　radicals，might　be　produced　frQm　the combustion　of　oxygen－containing　compomds.

For a　fault　detection　and　diagnosis　of　machinery，itis　essentialto　detecttlletypicalsignal　causedby　thefault．However， when　thesignal　would　bellotcontinuousbutintermittent，the　average level would below，and it became difficult to　detect it．Therefore，the　authors　highlighted　the　Wavelet　transform　as　one of　the　time-frequency　domain　analysis．This　paper concems　on　a　fault　detection　of　gear　damage　experimentaUy．To　model　the　damages，a　gear　tooth　was（or　gear　teeth　were）ground　out　on　its／their　face　or　broken　away．Wavelet　transform　was　applied　to　the　vibration　signal　received　by　a　sensor on　a　bearing　housing．As　the　result，the　time　series　data　of　magllitude　in　specified　frequencies　were　eamed．The　procedure is　considered　to　ludge　the　fault　by　that　data，Adding　to　this，this　method　is　compared　with　the　FFT（Fast　Fourier Trallsform）based　judgement．And　the　development　on　detectability　by　the　Wavelet　transfom　amlysis　were　clariHed.