home *** CD-ROM | disk | FTP | other *** search
- Comments: Gated by NETNEWS@AUVM.AMERICAN.EDU
- Path: sparky!uunet!paladin.american.edu!auvm!ASUACVAX.BITNET!AZPAB
- X-Envelope-to: stat-l@mcgill1.BITNET
- X-VMS-To: STATL
- X-VMS-Cc: AZPAB
- MIME-version: 1.0
- Content-transfer-encoding: 7BIT
- Message-ID: <01GOB5BON402002DZP@ACVAX.INRE.ASU.EDU>
- Newsgroups: bit.listserv.stat-l
- Date: Wed, 2 Sep 1992 12:40:45 -0700
- Sender: "STATISTICAL CONSULTING" <STAT-L@MCGILL1.BITNET>
- From: "Patrick Brooks" <AZPAB@ASUACVAX.BITNET>
- Subject: Fiber Distribution Problem
- Lines: 69
-
- To STAT-L folks:
-
- I am not sure if I should title this "The Distribution for Hell"
- or equivalently, "Somebodies Thesis" :-) but here goes:
- -----------------------------
-
- Do any of you have experience with characterizing distribution of
- diameters using Electron Microscopy? I am looking for
- literature, people, software that would help us. I will outline
- the problem in more detail below. I think this is a problem
- encountered in other physical sciences and hopefully you all will
- have some ideas.
-
- 1. A cross section of paper made of fibers (the fibers are
- roughly .2 to 10 microns in size) is exposed to Electron
- Microscopy. Particles of filler are also included. The
- assumption is that any thing over 10 microns in size is filler
- and ignored.
-
- 2. Two magnifications are used, 2000x and 4000x. Photographs are
- taken of the same region of sample. However, due to the
- magnification differences, the particles "seen" are different.
- There are 4 4000x photographs to 1 2000x photograph. But when
- you change magnification your "center" obviously changes. There
- is no way of making the 4000x region quadrant or known region of
- the 2000x magnification.
-
- 3. A computer program has the ability to calculate the "minimum
- diameter" for a closed edge that is completely in the view.
- Incomplete edges of fibers, that are on the edge of the view, are
- discarded.
-
- 3.1 PROBLEM TWIST: fibers are in different orientations. Thus
- the "minimum diameter" for each particle does not account for the
- angle of a fiber presented to the view. Fibers that are strait
- on will look like circles. Fibers at an angle will look like
- eclipses.
-
- 4. Bins are formed for "minimum diameter" intervals. There are
- more bins at 4000x than at 2000x. For a particular
- magnification, the computer counts the number of "minimum
- diameter" that fall into a bin.
-
- 4.1 WARM FUZZY: At 4000x accuracy/fuzzes is best between 0.2 and
- about 3 microns. With the warmest fuzzes being in the middle
- ;-). At 2000x, fuzzy is from about 1 micron to about 10.
-
- 4.2 PROBLEM: How can the data from the two magnifications be
- combined to describe the distribution of fiber diameter?
- 4.3 PROBLEM: How do you exclude by, by assumption, or include by
- theory :-), the orientation and fuzzes?
-
- 5. BIG PROBLEM: Given some type of empirical distribution can be
- found:
- 5.1 What is a good way to estimate central tendency, scale, and
- shape?
- 5.2 How can repeated measures from the same sample be combined?
- 5.3 Can comparisons from different samples be made?
-
- Thanks in advance
-
- Pat B.
-
- -----------------------------------------------------------------------
- Patrick Brooks | Of making many books there is no
- Bitnet: AZPAB@ASUACVAX | end, and much study is a weariness
- Internet: AZPAB@ACVAX.INRE.ASU.EDU | of the flesh.
- | -Qoheleth
- -----------------------------------------------------------------------
-