In the trace elements analysis, usually there are transient spikes caused by the "hiccup" of the machine. In iolite, is there any function that can identify and remove the spikes? This will help to correct the "wrong" signal produced by the machine and help to smooth the signals. In sills, it has such a function.

Thanks,
Shiqiang

    BTW, do you have function to choose two intervals for one sample?
    sometimes, there are some mineral inclusions in the samples, and I would like to avoid them. It would be better that I can choose two intervals for one sample so that I can get the enough signals.

    Shiqiang_Huang Hi Shiqiang,
    Yes, you can avoid spikes in iolite. In fact, iolite performs an outlier rejection automatically when reporting the mean and uncertainty for each selection. You can set the level of outlier rejection in Preferences -> Stats (2SD outlier, 3SD outlier, or none). By default, any data points greater than 3SD from the mean will be treated as outliers and not included in the reported mean for that selection.

    Even better is that iolite does this on a channel by channel basis, so that if a point is an outlier in one channel, it is not automatically then assumed to be an outlier in other channels. An example of when this is advantageous is if you have a signal spike which when one channel is ratioed to another is not an outlier (that is, the count rates might vary by more than 3SD, but the ratio between two channels for that datapoint is not an outlier, and iolite will not treat it as such).

    Regarding selecting two intervals for one sample (due to an inclusion or similar): yes, this is possible. It's called selection linking. You may have seen the little link icon that appears when you're editing one selection and then hover the mouse over another selection. If you click the link icon, the two selections will then create a linked selection. In the case of an inclusion, you could select the data before the inclusion, and then create a second selection for the data after the inclusion, and while you're still in Edit Mode, click the Link Icon on the first selection to link the two. iolite will report the stats for the combined datapoints as the Linked Selection. You can tell a that a selection is a linked selection because it is colored orange in the Time Series View.

    If you have any questions about these topics, please just let me know.

    Wow, this is great! I did not know that!
    Thanks very much. It is really helpful.
    BTW, I have some questions for you guys.
    In the lastest version of lolite, you guys have added some new data reduction schemes such as baseline substract and trace element next. For Baseline substract, it is easy to understand. But for the second one, I am kind of confused about this function. Could you tell me where I can find the information about this new feature?
    I really like the videos of the training short course you guys made last time about how to use iolite. I was wondering if you guys can make more short videos to show our customers what are the new developed functions and teach poeple how to use these functions and perhaps give some examples. This way, we can learn the new features from you guys more quickly. Reading is a good way to learn but using short videos are better than reading from my perspective. You guys can open a channel in youtube or somewhere putting all the training videos there. This can also help you guys to avoid answering the same questions again and again.

    Cheers,
    Shiqiang

    Hi Shiqiang,

    We'll be releasing more videos soon. We're just focussed on releasing a big new feature at the moment, but there will be more training/explanation videos coming out once it is released.

    Thanks for the feedback 🙂
    Bence

    Hi Bence,

    Thanks!
    I have one more question. When doing the data reduction of trace elements, does iolite incorporate the uncertainites of RMs and do the full error propagation?

    Best,
    Shiqiang

    Yes. This is the 'RMinc' uncertainty that you can export in your export settings.

    a year later

    Hi Bence,
    When doing the data reduction of U-Pb ages, how do we decide whether we should use error propagation to draw U-Pb plot and weighted average plot or not. I have watched the iolite U-Pb Data Processing Webinar, but I can't understand how it works.

    Thanks,
    Guochao

      Hi sunguochao

      I'd be happy to explain it in more detail if you'd like? Perhaps if you start a separate thread for it, as this thread is more about avoiding spikes.

      However, the short answer is that you should use the propagated error if you have calculated it.

      Kind regards,
      Bence

        Bence
        Thanks, Bence. I'd like to know more about the use of error propagation. The result seems reasonable when the error propagation is used. But I am not quite sure about the reason behind it.
        Cheers,
        Guochao

        Hi Guochao,

        I've just started a new discussion about the error propagation here.

        Please let me know if you still have any questions.

        Kind regards,
        Bence